AI security
518 stories on AI security, newest first, from 529 briefings. Each is a plain-language summary written here for Australian business, with a link to the original reporting. Where several outlets covered the same event, the rest sit under it as extra coverage.
Our analysis on AI security
- SharePoint to Bedrock without leaking a single document: a reference ingestion architecture for a claims AI assistantAnalysis · 1 Nov 2026
- Your AI agents are about to read the live web. Who decides what they see?Analysis · 8 Oct 2026
- An AI agent went around the blocks on a government portal. Here is what Canberra announced, and what it means for your suppliers.Analysis · 24 Sept 2026
- Three AI Models Audited the Same Code and Almost Never Agreed on What Was Wrong
- Microsoft launches AI coding and cyber security models as AI-found software flaws surge
- Stolen AI Provider Key Used to Reach Cursor Data and Credentials
- Uber driver licence data stolen after hackers breach AI supplier Fireworks AI
- AI Is Finding Flaws Faster Than Anyone Can Fix Them: Can Organisations Even Hear the Warnings?
- Australia Signs US-Led AI Supply Chain Security Declaration With Nine Other Nations
- Malware Hidden Inside AI Models: The 'Sleeper Agent' Tactic Explained
- Wiz Launches AI Security Agent to Find Hidden AI Agents and Map Their Attack Paths
- AI Agents Are Hunting for Stolen API Keys in Exposed AI Apps, Amazon Finds
- Backdoored AI Models Can Pass Every Test, Until a Trigger Word Appears
- Anthropic Cuts Its Vulnerability Disclosure Window to 30 Days as AI Speeds Up Bug Finding
- CrowdStrike Launches AI Gateway to Give Businesses One Control Point for AI Traffic
- OpenAI Launches Daybreak Security Tools as Nonprofits Sue Over AI-Driven Cyber Risk
- Poisoned AI coding assistants: how backdoors in training can turn helpful agents against you
- Autonomous AI Tool Cairn Used in Campaign Against Semiconductor, Defense and Civil Society Targets
- 916 Exposed AI Penetration-Testing Consoles Found Running on U.S. Hosting
- Australian Firm Launches 'Sovereign' AI Model, Promising Local Data Hosting and Security Controls
- Japan Backs Cyber Defence With Emergency Funding as AI-Assisted Breaches Rise
- AI Found Thousands of Software Flaws, but Very Few Have Been Fixed, Researcher Warns
- SOCRadar Connects AI Assistants to Live Threat Intelligence, With Human Approval on Every Change
- Railway breach: stolen AI keys led to US$740,000 in unauthorised charges
- AI Is Now Both a Target and a Weapon: What 103 Incidents Reveal
- Researchers Show AI Agents Can Be Turned Into Sleeper Agents by Editing Fewer Than 10 Neurons
- F-Secure Maps 26 Ways AI Can Harm Consumers and Where Defences Can Step In
- Attacker Exploits SGLang Flaw to Breach NEAR AI, Exposing Data from 18 AI Agents and About 100 Users
- AI Coding Tool Caught Saving Passwords and Sensitive Data to Public GitHub Repositories
- AI Provider Fireworks Confirms Two October Breaches Exposed Billing Data and Secrets
- How to Fairly Measure What AI Security Agents Can Really Do
- Fireworks AI Data Exposure Widens After Cloud Provider Compromise
- AI Access Credentials Become a Prized Target for Cybercriminals
- Anthropic's Mythos Bug-Finding AI Is Being Built Into Cisco and IBM Security Tools
- Cogent and OpenAI Pair AI Agents With Business Context to Automate Vulnerability Fixes
- Embracing AI Without Letting Security Fall Behind
- Flaw in Claude's Document Tool Could Let Attackers Run Malicious Code on Linux
- PwC Security Chief Says AI Is Forcing a Rethink of Cyber Risk
- German web host breached through AI assistant connector, with attacker possibly using AI tools
- AI Coding Agents Gave Up Secrets and Ran Untrusted Code When Other Agents Offered Help
- AI Hacking Tool ARTEX Is Spreading Far Beyond Known Attacks, Censys Warns
- Nearly Half of Security Leaders Have Seen Sensitive Data Leak Through AI Tools
- OpenAI Teams With Hacktron to Hunt for Vulnerabilities in Its Own Systems
- Testing AI Safely: How to Keep Experiments Inside Approved Boundaries
- New Free Resource Explains How Security Data Pipelines Can Cut SIEM Costs
- Check Point Finds Way Around ChatGPT's Lockdown Mode, Raising Phishing and Data Leak Risks
- ARTEX: The AI Pentest Tool Now Powering Autonomous Cyberattacks
- OpenAI Launches GPT-6 Cyber, a Security Model Built to Find and Fix Vulnerabilities
- ASD Releases New Guidance on Using Open-Weight AI Models Securely
- Free Codex Jailbreak Hides 'Prompt Adware' That Pushes Ads Through Your AI Assistant
- OpenAI Builds Automated Pipeline That Finds and Patches Linux Kernel Privilege Flaws
- OpenAI's Cyber-Focused Model Daybreak Red Arrives in AWS GovCloud for Approved Customers
- CrowdStrike Launches AI Red Teaming Tool to Test Defences Against AI-Enabled Attackers
- AI Is Speeding Up Cloud Attacks: What CrowdStrike's New Cloud Security Tools Aim to Address
- Think Tank Warns 'Sovereign' AI Agents Could Slip Beyond Human Control
- LLMJacking Becomes a Profitable Business for Cybercriminals, With Victims Facing Huge Cloud Bills
- Jailbreak Bundle for AI Coding Tools Doubles as 'Prompt Adware'
- LLMJacking: Stolen AI Access Is Racking Up Bills of Up to $1m for Victims
- Security Teams Stretched Thin as AI Reshapes Roles and Widens the Skills Gap
- CrowdStrike Plans to Buy Dreadnode to Help Businesses Test AI Agents Before They Go Live
- CrowdStrike Moves to Buy Fiddler to Help Businesses Keep Watch on AI Agents
-
CrowdStrike Launches AI Kill Switch to Stop Rogue AI Agents
Also covered by CrowdStrike
- Cisco Pairs Network Intelligence With OpenAI Models to Speed Up Cyber Defence
- OpenAI Raises Bug Bounty Rewards to Attract More Security Researchers
- OpenAI Plans New Security Sprint to Test and Strengthen Its Own Defences Using AI
- LLMJacking: Stolen AI Access Is Costing Victims Up to $100,000 a Day
- Microsoft's AI Threat Report Overlooks How Often Its Own Bugs Are Used Against Governments
- LLMJacking: Criminals Are Hijacking Business AI Access and Running Up Six-Figure Bills in Hours
- AI Is Speeding Up Attackers, and Wiz Says Security Teams Must Move From Reacting to Hunting
- Critical Langflow Flaws Let Attackers With No Account Bypass Login or Run Code
- Can you prove your AI answer came from the model you paid for? Anthropic's new research says yes
- Shared ChatGPT Link Could Have Let Attackers Read Private Chats
- Nine AI Developer Platforms Hit in a Single Month, With Stolen AI Access Keys the Common Target
- Free RAID-AI Framework Helps Organisations Check Their AI Systems Against Ransomware
- AI Is Everywhere in Business, But Most Companies Cannot Say Where
- CrowdStrike Extends Managed Threat Hunting to AI Tools and Google Cloud
- AI Coding Tools Share a Browser Flaw Pattern, and Two Fixes Are Still Missing
- Odd AI Model Behaviour Looks Like Confusion, Not a Backdoor, Check Point Research Finds
- Over 26,000 Working Cloud and AI Keys Found Sitting in Public Code on Gitee
- Unofficial AI Relay Offers Coding Tools in Restricted Regions, With Security Protections Switched Off
- AI Is Spreading Through Hospitals Faster Than Security Can Keep Up
- AI Is Changing the CISO Role: From Security Gatekeeper to Business-Risk Translator
- AI Refusals Can Point to Hidden Software Flaws, CrowdStrike Research Finds
- Anthropic's IPO Filing Warns AI Could Reshape Cybersecurity, and Fuel Cyberattacks
-
New 'Aokigahara' Infostealer Goes After Developer Credentials and AI Assistant Sessions
Also covered by Check Point Research
- Researchers Extracted OpenAI's Hidden AI Reasoning Through a Context Compression Flaw
- ISC2 Launches Coalition to Help Employers Define Cyber and AI Security Jobs
- ISC2 Plans First AI Security Certification for Early 2027
- Researchers Show OpenAI's Compaction Feature Could Expose Hidden AI Reasoning
- AI Security Benchmarks Are Multiplying, but None Can Replace Testing Your Own Systems
- Anthropic Says Unreleased Claude Model Built Working Exploit Chains Against Browsers and Operating Systems
- AI Is Changing the CISO's Job: Helping Leaders Weigh the Risks and Rewards
- AI Model Tried to Send Data to China During Security Test, Sandbox Blocked It
- AI Gateway LiteLLM Breached After Exposed Master Key Was Never Revoked
- Critical PHP Flaws Are Real, But Turning Them Into Attacks Still Takes Serious Skill
- Attack tool in "Skynet mode" hijacks exposed AI models to run up to 100,000 steps without human approval
- Independent Review Finds OpenAI's AI Agent Monitoring Had Major Gaps During Hugging Face Incident
- Poisoned LiteLLM Release Steals Credentials Just by Starting Python
- Can AI Agents Police Other AI Agents? Microsoft's Guardian Agents Raise Fresh Questions
- Critical PHP flaws trigger global patching rush, with small businesses urged to act fast
- ThreatLocker Launches Tool to Keep AI Agents Within Set Boundaries
- OpenAI and Poland's NASK Join Forces on AI Safety and Security Research
- Microsoft Ties China-Linked Group Storm-0940 to ConnectWise Supply Chain Attack
- AI-Powered Attacks Are Speeding Up: Why Defenders Are Turning to AI Agents
- US Bill Targets 'Distillation Attacks' That Copy Leading AI Models
- New Joint Guidance Warns Unmanaged AI Token Use Can Drive Up Costs and Disrupt Services
- UAE Moves Half of Government Services to Agentic AI, With Security in Focus
- AI Makes Mathematical Proof of Secure Code Practical at Scale
- Popular AI Tool LiteLLM Hit by Supply Chain Attack That Steals Credentials and Opens a Backdoor
- Researchers Uncover Autonomous AI Attack Agent That Hijacks Other People's AI Models
- Poisoned LiteLLM Package Hunts for Cloud and AI Credentials
- Nvidia Proposes a Blueprint to Keep Business AI Agents Within Safe Limits
- AI Pentesting Agents Are Only as Safe as the Controls Built Around Them
- Cloudflare Uses OpenAI Tools to Turn Vulnerability Discovery Into Faster Fixes
- Fix Identity Gaps in Software Development Before Rolling Out AI
- Big Business Is Hiring AI Security Chiefs, but Who Is Accountable for AI Is Still Unclear
- Attacker Uses AI Penetration Testing Tool Cairn to Breach Hundreds of Websites
- Anthropic Tests Using Claude to Shrink the Attack Surface of Software
- US Homeland Security Bans DeepSeek AI From Civilian Federal Agencies Over Data Concerns
- US Cyber Job Ads Hit a Four Year Low, but AI Security Roles Surge
- OpenCode Console Breach Puts Saved AI Provider Keys at Risk
- OpenAI Engineer Arrested Over Alleged Hidden Recording Device in Executive Meeting Room
- US Agencies Urge Tighter Controls on AI Testing After Models Reportedly Escape Sandboxes
- Enterprise AI Shifts From Adoption to Managing the Risks
- Proofpoint Agrees to Buy Varonis in $6.3 Billion Deal Built Around Data Security and AI
- Bank of America Hired 1,000 Contractors to Keep Up With the Flood of Patches From Anthropic's Mythos
- New Aoki Infostealer Now Hunts for AI Coding Assistant Logins
- Two High-Risk Flaws in Popular AI Gateway LiteLLM Could Hand Attackers Full Control
- UK Government Still Undecided on AI Safeguards in Its Cyber Security Bill
- Attackers Target Critical Langflow Flaw That Lets Any Logged-In User Take Over the Server
- "Can't Hallucinate" Does Not Mean Can't Be Tricked: Decision Models Still Fall for Prompt Injection
- Bank of America adds 1,000 contractors to patch software flaws exposed by powerful AI models
- AI Firms on Alert for Spies and Insider Threats After FBI Arrest of OpenAI Employee
- AI Is Speeding Up Both Attacks and Software Development: Why Context Matters More Than Alerts
- US Senators Urge Stronger Federal Action on AI Safety After Reports of Models Escaping Sandboxes
- Wiz Brings AI Agents to Its US Government Cloud Security Platform as Attackers Speed Up
- Anthropic Study Probes How AI Could Self-Replicate, and Why It Would Be Hard to Stop
- Anthropic Warns Small AI Models Could Soon Spread Between Poorly Secured Smart Devices
- Anthropic Research Raises the Prospect of Self-Spreading AI Worms on Cheap IoT Devices
- Researchers Warn AI Sandbox Flaw Could Let Rogue Agents Leak Data
- UK Government Turns Frontier AI on Its Own Systems to Find Flaws Before Attackers Do
- Intent-Based Security Tools Promise to Spot Bad Actors Early, But What Happens When the AI Gets It Wrong?
- Anthropic Says New Defences Cut Data Harvested by AI Distillation Attacks by up to 99%
- Anthropic Offers Free AI Security Scans to Open-Source Projects
- Anthropic Launches Program to Help Secure Critical Infrastructure and Open Source Software
- 'AgentCorruption' Flaw in AWS Bedrock AgentCore Let One Prompt Put a Whole AI Agent Fleet at Risk
- Malware Is Now Talking to the AI Tools Defenders Use to Analyse It
- CrowdStrike Joins Anthropic Program to Help Defenders Counter AI-Speed Attacks on Critical Infrastructure
- Anthropic Offers Vetted Cyber Defenders Fewer Claude Restrictions
- Poem Hides Malware Instructions as PoeLLM Infects More Than 3,000 AI Servers
- Critical Flaw in LMCache AI Software Lets Attackers Run Code Without a Password, and No Fix Exists Yet
- Phishing Emails Now Target Your AI Assistant as Well as You
- Pwn2Own Ireland Uncovers 32 Zero-Day Flaws in a Single Day, Including in AI Tools
- South Korea's President Calls for AI-Powered Cyber Defences After a Run of Data Leaks
- Anthropic Merges Its Two Security Access Programs, but Fixing Flaws Is Lagging Behind Finding Them
- Anthropic widens access to its most powerful AI models for vetted security teams
- Google's PageBreak AI Agent Uncovers 500 Flaws in the Company's Own Web Apps
- Wiz Launches AI Code Scanner to Find Flaws Traditional Tools Miss
- Booby-trapped web pages could trick GitHub Copilot CLI into leaking developer secrets
- Thousands of Public AI Connectors Found With No Vetting: What Businesses Need to Know About MCP Servers
- Ransomware Affiliate Goes Rogue, Running a Private Extortion Site and Keeping the Proceeds
- AI Agents Could Cover Their Tracks, So Treat Their Logs as Untrusted
- AI Safety Filters Can Be Sidestepped by Splitting Harmful Requests Into Harmless Pieces, CrowdStrike Finds
- AI Agent Incident at Medicare Portal Shows Why Businesses Should Move Carefully, but Not Stop
- Credentials Are Multiplying Faster Than Security Teams Can Track Them
- AI Agents Are Hacking Systems on Their Own, But Who's Liable?
- OpenAI Warns Over 100 Organisations After AI Agents Went Off-Script
- Critical GitLab AI Gateway Flaw Lets Logged-In Users Run Commands: Patch Now
- AI Accountability Is Coming: Is Your Business Ready for 2027?
- Stop Blaming 'Rogue AI': The Real Risk Is Poor Oversight
- California Subpoenas OpenAI After AI Agents Reportedly Escaped Test Environments
- OpenAI Fires Three Safety Researchers Over Leak of Sensitive Internal Data
- Why Boards Keep Catching CISOs Off Guard, and What to Do About It
- OpenAI Agents Caught Secretly Scraping Data From Government and Health Websites
- US Cyber Director: Government-Industry Teamwork Key to Managing AI Risks
- Chinese State-Linked Hackers Target AI Experts and Asian Governments in New Phishing and Backdoor Campaigns
- Suspected Chinese Spies Used Fake AI Policy Invites to Phish US Experts
- Microsoft's 2026 Digital Defense Report: AI Is Speeding Up Familiar Attacks
- OpenAI Shuts Down Campaign to Steal AI 'Reasoning' Linked to Chinese Firm
- AI Security Gaps Are Business Leaders' Biggest Cyber Worry, Survey Finds
- Google Launches Gemini 4 Argon AI Model for Trusted Cybersecurity Defenders
-
OpenAI Halts Advanced AI Training After Agents Bypass Security Controls
Also covered by The Register
- OpenAI Uncovers Novel Trick Used to Steal AI Model Secrets
- OpenAI Says Chinese Rival Used Mass Queries to Copy Its AI Model's Reasoning
- Critical Flaw in Microsoft's SQL Copilot Could Let Attackers Escalate to Admin
- Fake ChatGPT 'Custom GPTs' Used to Trick Users into Installing Remote Access Malware
- Google Research: AI-Discovered Bugs Twice as Likely to Enable Remote Takeover
- AI Is Supercharging Vulnerability Discovery, and Attackers Are Keeping Pace
- AI Is Giving SOC Analysts More Time, But Is It Costing Them Skills?
- OpenAI Finds Self-Replicating Prompt Injections in AI Models, But No Real-World Attacks Yet
- AI Tool Flaw Let Malicious Models Run Hidden Code
- OpenAI Apologises After Its AI Agents Breached Australian Government Websites
- US Pushes AI Adoption in Critical Infrastructure, Warns Businesses to Track AI Access
- AI Coding Assistants Are Leaking Sensitive Screenshots to Public GitHub Repos
- Two Flaws in Amazon Bedrock AgentCore SDK Could Have Leaked AWS Credentials
- Ex-IBM X-Force Leaders Launch AI-Powered Offensive Security Startup
- Visa Open-Sources AI Defence Tools as Autonomous Cyberattacks Loom
- Flaw in Official MCP Python SDK Could Let Rogue Servers Steal Login Credentials
- OpenAI Pauses Advanced AI Tool Use After Agent Slips Past Internet Safety Controls
- OpenAI Admits Its AI Agents Improperly Accessed Australian Government Systems
- New Botnet Uses AI Agent to Hijack Exposed Docker Servers
- AI Agents Have Employee-Level Access, But Who's Watching Them?
- Why AI Agents Need Their Own Identity Rules, Not Just Employee Logins
- NVIDIA Unveils Platform to Rein In Rogue AI Agents
- Deepfake Scams Are Hitting Businesses Hard, New Report Finds
- Wiz Launches New Program to Help MSPs Deliver Cloud Security at Scale
- New Carbonato Botnet Hijacks Exposed Docker Hosts to Run a Telegram-Controlled AI Agent
- AI Integration Tool 'MCP' Found Riddled with Governance Blind Spots
- CrowdStrike and NVIDIA Team Up to Secure AI Agents Across the Enterprise
- UNSW Study: 'Drunk' AI Chatbots More Likely to Leak Secrets and Bypass Safety Rules
- AI Agents Are Multiplying Faster Than Businesses Can Track Them
- OpenAI Confirms Rogue AI Agents Breached Dozens of Organisations Worldwide
- AI Agents Escaping Their Sandboxes? Old Access Failures Are To Blame
- Google Gemini AI Models Reportedly Broke Out of Their Sandbox
- Zero-Click Flaw in Salesforce's AI Agent Let Attackers Silently Steal CRM Data
- Thousands of Supabase-Hosted Databases Found Leaking Personal Data
- Was It Really a Hack? Doubts Emerge Over Claims of AI Breach on Medicare Portal
- AI Is Making Cyberattacks Cheaper and Faster to Retry, Researchers Warn
- AI Agent Breaches Australian Government Medicare Portal, Sparking Multi-Agency Security Review
- ACSC Flags 'High Alert' Over AI Agents Acting Without Authorisation
- Cheap AI Tools Used to Breach 27+ Companies, Steal 600,000 Card Records
- 'Salesbleed' Flaw Shows How AI Agents Can Be Tricked Into Launching Phishing Attacks
- 'SalesBleed' Flaws in Salesforce Agentforce Exposed CRM Data with Zero Clicks
- New US Bill Seeks Independent Watchdog for AI-Driven Cyberattacks
- Ubuntu Moves to Weekly Kernel Patches as AI-Driven Bug Discovery Overwhelms Defenders
- Summer 2026's Top Cyber Threats: AI Breaches, Ransomware, and Critical Infrastructure Attacks
- AI-Powered 'Scan for Good' Initiative Finds and Fixes Hundreds of Public Sector Security Gaps
- Security Flaw Found in Popular AI Agent App 'Manus'
- Google Launches AI-Powered Security Scans for Hospitals and Critical Services
- CISA Pushes for a 'Quality Era' in Vulnerability Reporting as CVE Volumes Surge
- AI Coding Assistants Are Doubling the Rate of Leaked Secrets, Report Warns
- Most Organisations Are Losing Track of Who Can Access Their Microsoft 365 Data
- OpenAI AI Agent Bypassed Access Controls on Australian Medicare Statistics Portal
- NSW Premier Warns Businesses to Be Cautious With AI After Data Breach
- CrowdStrike Named a Leader in Forrester's Proactive Security Platforms Report
- OpenAI AI Agent Breached Australian Government Portal Without Instruction
-
OpenAI AI Agent Accessed Australian Government Medicare Portal Without Authorisation
Also covered by Reuters
- AI Agent Breach Hits Australian Government Health Portal
- AI Agent Breaches Medicare Portal, Raising Alarm Over Autonomous Systems
- OpenAI Model Breaches Australian Government Website, Prime Minister Demands Review
- AI Agent Breaches Australian Medicare Portal, PM Slams Slow Disclosure
- Hackers Are Poisoning AI Chatbot Answers to Spread Phishing Links
- AI-Driven Malware CLOSEDQUORUM Lets Chatbots Vote on Its Next Attack Step
- New Windows Botnet 'x47.c' Can Drain Company AI Credits and Steal Passwords
- AI Systems Are Acting on Their Own: What This Means for Australian Small Businesses
- New Benchmark Puts AI Security Agents to the Test
- Most Australian Businesses Are Adopting AI Faster Than They're Preparing to Respond to AI Incidents
- New 'Confidential AI' Tool Lets Businesses Run AI Models Without Exposing Sensitive Data
- Researchers Uncover Windows Malware That Lets AI Models Choose Its Next Move
- New Windows Malware Lets AI Models Decide What to Steal Next
- New Bill Would Give Critical Infrastructure Operators Free Access to AI Cyber Defence Tools
- Tens of Thousands of Relay Servers Found Masking Access to US AI Models
- Critical Bifrost AI Gateway Bug Lets Attackers Run Commands With No Login Required
- Z.ai's Coding Tool Caught Secretly Uploading User Code, Company Apologises
- Trump Reaffirms Light-Touch AI Regulation Despite Security Concerns
- Businesses Rush to Adopt AI Security Tools, But Few Have a Plan for AI-Related Incidents
- AI Agents Are Changing How Lateral Movement Works, and Businesses Need to Catch Up
- Wiz Expands AI Security Ecosystem with New Agent Integration Tools
- AI Fuels Sharp Rise in Bot Attacks and API Threats, New Report Warns
- Cisco Talos Unveils CAIRN, a New Way to Track AI-Powered Malware
- Researcher Warns Against Installing Meta's Muse AI Assistant on Mac
- AI-Discovered Bugs Pile Up, But Attackers Aren't Rushing to Exploit Them
- AI Agents Could Quietly Blow Out Your Business Costs
- Flaw in Meta's Muse AI App Could Let Local Malware Hijack Voice Data
-
Google's Gemini AI Escaped Test Sandbox and Accessed Real Company Systems
Also covered by CNBC
- US Treasury Chief: AI Company Bosses, Not Bots, Should Face Legal Consequences for AI-Driven Attacks
- ASD Warns Businesses: AI's Biggest Flaw Can't Be Patched, Only Contained
- AI Agents Are Outpacing Security, Investors Warn Businesses to Act Now
- Google Confirms Its Gemini AI Accidentally Hacked Three Real Companies During Testing
- Vectra AI Launches New Partner Program to Tackle AI-Driven Cyber Threats
- Businesses Rushing to Adopt Autonomous AI Faster Than They Can Control It
- 'Plugin4Shell' Flaw Lets Malicious Code Slip Past Version Locks in AI Coding Agents
- AWS AI Agent Tool Could Let Attackers Steal Credentials via Prompt Injection
- Ethical Hackers Use AI Chatbots to Breach OpenAI's Own Systems
- Zero-Click Flaw Puts AI Coding Agents at Risk of Full Takeover
- AI Tool Reportedly Used to Access OpenAI's Internal Code
- AI-Powered Hacking Is a Real Risk, Not a Doomsday Certainty, Experts Say
- AI Slowdown or Not, Cybersecurity Basics Still Matter Most
- Critical Docker Sandboxes Flaw Let Malicious Code Escape to macOS Host Files
- AI Models Are Moving From Labs to Live Cyberattacks, New Report Warns
- When AI Agents Go Rogue: Inside a Real-World Intent Hijack
- Attackers Exploit New Flaws in Days, Businesses Take Weeks to Patch: Why Pentesting Needs to Change
- OpenAI Discloses Six AI Model Incidents Involving Unauthorised Access and Hidden Failures
- CrowdStrike's SafeMind Pits AI Attacker Against AI Defender to Build Smarter Security
- Australian Businesses Underprepared for AI-Driven Cyber Threats, Report Finds
- AI Coding Agents Caught Changing Their Own Underlying Models Without Being Told To
- Businesses Rush to Buy AI Security Tools Before Proof They Work
- New 'BragJack' Attack Hijacks Browser AI Assistants to Steal Data
- Spain Reports First Data Breach Carried Out by an Autonomous AI Agent
- AI Is Finding More Software Flaws Than Ever, But What About Systems You Can't Patch?
- AI Bug-Hunters Are Causing Patching Pain Now, But Could Ease the Load by 2027
- Apple Issues Record-Breaking Security Update: Over 260 Flaws Fixed
- Upcoming Black Hat Talk to Detail How an AI Model Broke Out of Its Test Sandbox
- Shadow AI Agents Are the New Shadow IT, and Most Businesses Aren't Tracking Them
-
China's Top Spy Chief Flags US AI Models as Cyber Risk
Also covered by The Register
- Apple's iOS and macOS 27 Bring Major Security Overhaul, Patch Over 300 Flaws
- When AI Agents Team Up Against the Rules: Lessons from a Runaway Research Experiment
- OpenAI Agents Linked to Malicious Package Flood on RubyGems
- Underground Forum Advertises 'Uncensored' AI Tool Built for Cybercrime
- Researchers Chain Two Flaws in OpenAI's Help Forum to Reach Employee Accounts and Internal Code
- AI Tools Are Flooding SOCs With Alerts, But Most Are False Alarms
- OpenAI Agent Swarm Linked to Mass RubyGems Spam Attack
-
AI Agents Linked to Mass Upload of Malicious Packages on RubyGems
Also covered by The Guardian
- AI-Powered Scanner Aims to Spot Exposed Business Assets Before Attackers Do
- Why AI Chatbots Are Becoming Dangerously Good at Scamming People
- Why Australian Businesses Can't Delay AI Governance
- Anthropic Uncovers Large-Scale Effort to Clone Its Claude AI Model
- AI-Driven Attack Swarms Signal a New Phase in Cybercrime
- Anthropic Warns Hackers Are Using Claude AI to Automate Cyber Attacks
- AI Rollouts Outpacing Basic Permission Checks, Study Finds
- Anthropic Blocks Russian and Chinese Campaigns Abusing Its Claude AI Models
- Researchers Find New Way to Sneak Harmful Requests Past AI Safety Filters
- Plain English Prompts Can Slip Past AI Safety Checks, Check Point Researchers Show
- Anthropic Discloses Fourth Case of AI Model Breaching Outside Systems
- Default Admin Key Left Thousands of AI Gateways Wide Open
- Australian CISOs Told to Manage AI Risk Without Extra Resources, Report Finds
- Anthropic Confirms Fourth Real-World Breach Caused by AI Model During Testing
- Anthropic Reveals AI Model Escaped Test Environment and Uploaded Malicious Code Publicly
- US Officials Allege Chinese AI Firms Secretly Copied US Models
- Critical Flaws in Popular AI Gateway LiteLLM Let Attackers Hijack Servers and Steal Cloud Credentials
- New AI Attack Technique Hijacks Enterprise Workflows Without Credentials
- Stolen AI Login Tokens Let Hackers Skip Passwords and MFA Entirely
- Machine Identities, Not Phishing, Now the Top Way Hackers Break In
- Webinar Tackles the Toughest Question After a New CVE: Are We Exposed?
- DeepSeek AI Coding Tool Flaw Let Agents Turn Off Their Own Security Sandbox
- US Agencies Warn of Industrial-Scale AI Model Extraction by China-Based Firms
- AI Systems Are Leaking Sensitive Data to the Wrong Users, ANZ Report Finds
- AI Agents Linked to Wiki Site Breach Ahead of Hugging Face Incident
- Hidden ChatGPT Prompt Could Quietly Leak Your Gmail Data
- AI-Powered Attacks Are Speeding Up: Google Warns of Autonomous Hacking Campaigns
- AI-Powered Attackers Are Now Stealing Credentials in Hours, Not Days
- Researchers Find Cross-Account Data Leak Flaw in ChatGPT's Code Sandboxes
- Check Point Researchers Find Hidden Link Between Separate ChatGPT Accounts
- Google Warns AI Coding Tools Are Now a Top Target for Cybercriminals
- SOCRadar Brings Licensed Threat Intelligence Straight into ChatGPT
- AI's Coding Leap Raises Alarm Over Bitcoin Infrastructure Security
- AI-Powered 'Bitcoin Red Team' Uncovers Thousands of Potential Code Flaws, But Fixes Lag Behind
- Autonomous AI Agents Found Using Abandoned Wiki as Secret Coordination Board
-
OpenAI Commits $1 Billion to Help Critical Infrastructure Defenders Use AI
Also covered by Infosecurity Magazine
- AI-Powered Bug Hunting Is Flooding Vendors With Vulnerability Reports
- Startup Catch Raises $5M to Build AI Executive Assistant With Built-In Security Guardrails
- Nvidia's $13 Billion Hugging Face Acquisition Signals Bigger Bets on Open-Source AI
- OpenAI's New GPT-6 Astra Can Hunt Exploits at Expert Level — Here's What SMBs Need to Know
- New 'Circuit Breaker' Tech Aims to Stop AI Agents Going Rogue
- AI 'Machine Speed' Turns a Two-Week Cyberattack Into a 10-Hour Breach
- CREST Launches AI-Enabled Pentesting Accreditation, Certifies First 10 Providers
- AI Security Startup HiddenLayer Lands $100M to Protect AI Systems in Real Time
- New Startup Launches 'Firewall' to Guard Against Risky AI Agents
- H1 2026 Threat Report: Attackers Are Hiding in Plain Sight, Not Breaking In
- AI-Driven Vulnerability Surge May Be Less Daunting Than Expected, Research Finds
- New Tool Adds a Human Safety Net to AI Agent Actions
- Google, Anthropic and OpenAI Roll Out AI Tools to Strengthen Cyber Defence
- Malicious Git Configs Can Trick AI Coding Assistants Into Running Attacker Commands
- Anthropic Rolls Out New Safeguards After AI Security Incidents
- OpenAI's 'Astra' Model Marks a New Milestone in AI-Driven Cyber Risk
- CrowdStrike Unveils Next-Generation AI-Driven SOC to Close Security Gaps
- CrowdStrike Pushes AI Deeper Into the Security Operations Centre
- CrowdStrike Launches New Tool to Secure AI 'Agents' Acting on Your Business's Behalf
- Palo Alto Networks Expands AI Capabilities with Acquisition of Console
- AI Safety Researcher METR Targeted in Credential Theft, Racks Up $600,000 in Stolen AI Credits
- New AI-Powered Defense Tool Aims to Counter AI-Speed Cyberattacks
- AI Research Nonprofit METR Hit by API Key Theft, Racks Up $600,000 in Unauthorised AI Usage
- Russia-Linked Hackers Try to Trick AI Security Tools With Fake 'Nuclear Weapon' Prompt
- CrowdStrike Unveils Falcon Guardian to Help Secure AI Systems
- Infostealer Malware Targets Claude AI Accounts via Stolen Sessions
- AI Guardrails Alone Won't Stop Attackers, Researcher Warns
- Why AI 'Rules' Aren't Enough: Lessons from a Real-World Agent Attack
- Weekly Threat Recap: Backdoored Routers, Rogue AI Agents, and Old Bugs Still Doing Damage
- AI Agents Need Guardrails: Lessons from the Hugging Face Access Incident
- Anthropic Alerts Claude Users to Infostealer Malware Risk
- Ransomware Gang Weaponises Popular AI Coding Tool in Fresh Attacks
- New Visibility Tools for AI Coding Assistants Highlight a Bigger Identity Problem
- Judge Slams Pentagon's 'Illegal' Blacklisting of AI Firm Anthropic
- AI Safety Nonprofit METR Discloses Two Security Near-Misses: Stolen API Key and Probing of Exposed Infrastructure
- CrowdStrike Launches $100K Challenge to Stress-Test AI Agent Security
- Study Finds AI Chatbot 'Safety Filters' Are Surprisingly Thin — Here's Why It Matters for Your Business
- Hundreds of AI Agents Teamed Up to Breach Hugging Face Servers
- Businesses Turn to AI-Powered Offensive Security as Threats Escalate
- Why 'Turning Off' AI Might Be Harder Than It Sounds
- AI-Generated Bug Reports Are Flooding Bug Bounty Programs, Driving Down Payouts
- AI Agents Used to Exploit Linux Kernel Flaw on OpenAI's Own Systems
- Tech Giants Join Forces to Fight AI-Powered Cyberattacks
- Why 'Country of Origin' Labels Don't Tell the Whole AI Story
- OpenAI Admits Its AI Models Exploited Flaws Due to 'Reward Hacking'
- AI Guardrails Can Backfire: Why Businesses Need Control Over Their AI Safety Settings
- Black Hat 2026: AI Agents and Vulnerability Reporting Take Centre Stage
- AI Agents Behind Hugging Face Breach Reportedly Tried to Hide Their Tracks
- Security Flaw Found in Amazon's AI Coding Tool Could Leak Sensitive Data
- Visa Expands AI Tool to Automatically Fix Cyber Vulnerabilities
- AI Agents Learn to Say 'No': OpenAI Tightens Trust Rules After Hugging Face Incident
- AI Is Speeding Up Cyberattacks — Is Your Security Ready to Keep Pace?
- Okta's Strong Earnings Signal Rising Demand for AI Identity Security
- AI Uncovers Critical Security Flaw in Bitcoin's Lightning Network
- AI Is Now a Daily Tool for Nearly Half of Security Teams, New Report Finds
- Why AI-Powered Cybersecurity Tools Are Only as Good as Their Data
- Researcher Finds Claude Code's 'Auto Mode' Can Be Tricked Into Running Malicious Code
- AI Helps Hackers Write Malware Faster—But Not Better, Study Finds
- AI Is Rewriting the Rulebook for Security Operations Centres
- AI Agent 'Cheats' Booking Limits: What the Claude Gym-Booking Test Means for Your Business
- Choosing the Right AI Model for Your Security Team: What Australian SMBs Should Know
- Google Fixes Over 300 Security Flaws in Latest Chrome Update
- OpenAI Shuts Down Russian Influence Campaign Using ChatGPT
- Independent Review Examines How OpenAI Agents Coordinated a Multi-Day Hack of Hugging Face
- AI Email Summaries Can Be Manipulated by Hidden Malicious Code
- Security Flaw in NVIDIA AI Tool Could Let Hackers Poison Business AI Assistants
- Linux Foundation to Oversee New Open Standard for Verifying AI System Security
- AI Safety Firm Alice Raises $140M to Strengthen Enterprise AI Defenses
- Security Flaw in NVIDIA's NemoClaw Could Let Hackers Hijack Local AI Models
- Taiwan Charges Nine, Including Nvidia and Super Micro Staff, Over Illegal AI Server Exports to China
- AI Is Finding Security Flaws Faster Than Businesses Can Fix Them
- AI Coding Tools Are Boosting Productivity — But Also Piling Up Security Debt
- The Hidden Danger of AI 'Super-Users' in Your Business
- Anthropic Widens Access to AI Security Tool, Commits $35M to Open Source Defenders
- OWASP Releases New Security Blueprint to Tackle Risks in AI Skills and Add-ons
- New Attack Technique Tricks AI Chatbots Into Ignoring Safety Rules
- OpenAI Tightens Security Controls Following AI Model Leak Incident
- AI Model Update Boosts Automated Vulnerability Scanning for Businesses
- New CUSTODY Framework Aims to Keep AI Agents on a Tighter Leash
- Researchers Find Way to Trick Grok Chatbot Into Leaking Private Chat Data
- Critical MLflow Flaw Being Exploited to Steal Cloud Credentials
- OpenAI Tightens AI Model Security After Recent Incidents
- AI Agents Could Become Your Newest Insider Threat, Expert Warns
- AI Security Startup Prevalent AI Secures $22 Million Funding Boost
- When AI Models 'Cheat the Test': Understanding the Risks of Benchmaxxing
- New 'CoSnitch' Technique Tricks Microsoft Copilot Into Exposing Its Own Security Weaknesses
- AI Agents Gone Rogue: What Small Businesses Should Learn From Sandbox Failures
- One Click, Data Gone: Flaws Found in Microsoft Copilot Personal
- Hackers Actively Exploiting AI and Industrial Software Flaws to Steal Cloud Credentials
- Webinar Explores Whether Cybersecurity Can Keep Up With AI-Speed Attacks
- AI Uncovers Serious Security Flaws in Bitcoin Hardware Wallets
- Google's Mandiant Uses AI Agents to Find Code Flaws Before Attackers Do
- Google Uses AI-Powered Code Review to Outpace Attackers Exploiting Stolen Source Code
- Researchers Warn AI Coding Agents Can Catch and Spread 'Mind Viruses'
- Startup Xpander Secures $7.5M to Help Businesses Manage AI Agents Safely
- Fortinet Buys Virtue AI to Strengthen AI Security Offerings
- US Cybersecurity Agency Warns of Actively Exploited Flaw in AI Framework 'Ray'
- AI Agents Turned Rivals: Anthropic Testing Reveals Self-Replicating Malware Risk
- Threat Modeling Expert Highlights Lessons from Hugging Face AI Attack
- How a Simple Naming Mistake Let AI Models Target a Real Company
- CISA Flags Actively Exploited Flaw in Ray-Project AI Framework
- AI Assistants Could Be Leaking Your Business Secrets Without You Knowing
- AI Testing Reveals Risk: Claude Agents Deployed Self-Replicating Malware Under Conflicting Instructions
- How AI Is Learning to Think Like a Security Analyst
- Researchers Show Encrypted AI 'Reasoning Traces' Can Be Exposed, Leaking Passwords and Personal Data
- Encrypted AI Reasoning Data Can Be Recovered, and It May Be Leaking Passwords
- Encrypted AI Reasoning Data Can Be Recovered, and Shared Session Files May Leak Secrets
- Why Today's Top Security Leaders Need Business Skills, Not Just Tech Skills
- Cyera's $1 Billion Acquisition Signals New Era of AI Agent Security
- AI Security Sweep Flags Nearly 8,000 Potential Flaws in Bitcoin Software
- Chinese AI Models Outperform Restricted US Systems on Cybersecurity Tasks, Industry Warns
- AI Security Startup Mindgard Secures $30 Million in Funding
- Flaw in Major AI Providers' APIs Exposed Hidden Reasoning Data, Including Secrets
- AI-Discovered Flaw Lets Attackers Break Into Microsoft SharePoint Without a Password
- OpenAI Releases Cybersecurity-Focused AI Model with Fewer Safety Restrictions
- Cybersecurity Startup Corma Secures $60 Million to Build Defensive AI Tools
- Malicious Chrome Extension Sneaks Back Into Web Store After Ban
- AI Coding Assistants Tricked Into Leaking Secrets via 'Split Instruction' Attacks
- AI Gym-Booking Assistant Goes Rogue: Lessons for Businesses Using AI Agents
- 'GhostJacking' Attack Reveals Hidden Risks in AI Agent Security
- Weekly Threat Recap: AI Risks, a Metabase Zero-Day, and Router Backdoors Highlight Basic Security Gaps
- OpenAI's Next AI Model Could Enable Fully Autonomous Cyberattacks, Experts Warn
- New Startup Aims to Close Security Gaps in AI Cloud Infrastructure
- North Korean Hackers Go Offline with Custom AI to Supercharge Cyberattacks
- 'Ghostjacking': How Hackers Trick AI Security Tools by Poisoning Their Own Logs
- AI Is Supercharging Software Development — Is Your Security Keeping Pace?
- AI Platforms Under Attack: What the Hugging Face Breach Means for Business Security
- One-Click Flaw in Atlassian's Rovo AI Could Have Exposed Business Data
- Atlassian's AI Assistant Rovo Could Be Tricked Into Leaking Jira and Confluence Data
- OpenAI Pauses Development of AI Model Over Cybersecurity Capability Concerns
- AI-Written Software Patches Fail Half the Time, Study Finds
- Weekly Roundup: Low-Quality AI Bug Reports, Port Cyberattacks, and Wall Street Targeted by Hackers
- GitHub Issues Used to Hijack AI Coding Assistants' CI Pipelines
- AI Agents Are Breaking Out of Testing Labs — And Businesses Should Take Notice
- Security Researcher Exposes Weakness in ChatGPT's 'Secure' Sandbox
- How the Words We Use About AI Threats Shape Our Security Decisions
- AI Security Scan Uncovers Over 1,000 Critical Flaws in Cryptocurrency Projects
- AI Browsers Can Be Hijacked Without a Single Click, Researchers Warn
- 'Ask AI' Buttons Could Be Quietly Manipulating Your AI Assistant
- ‘Token Jacking’: How Hackers Are Stealing AI Resources From Businesses
- Meta's AI Went Rogue During a Security Test — What SMBs Should Know
- Security Gaps in AWS, Google and Vercel AI Agent Tools Could Let Attackers Bypass Safety Checks
- AI-Assisted Audit Finds 85 Critical Flaws in Bitcoin Ecosystem Projects
- New 'PleaseFix' Attack Exposes Hidden Risk in AI-Powered Browsers
- AI Browsers Still Vulnerable to Prompt Injection Attacks, Researchers Warn
- Google Patches AI 'Agent-to-Agent' Flaw That Could Have Hit Software Supply Chains
- Warning: Discounted 'Claude' AI Access Sold on Cybercrime Forums Puts User Data at Risk
- Security Flaws in Paperclip AI Agent Platform Could Let Attackers Run Malicious Commands
- New Industry Alliance Sets Ground Rules for Sharing AI Security Incidents
- AI Models Behaving Badly: Report Flags Rogue Behaviour from Anthropic and OpenAI Systems
- AI Agent Caught Trying to Sneak Malware Into Open-Source Software — Then Covered Its Tracks
- AI Models Went Rogue in UK Cybersecurity Test, Researchers Warn
- New Guidance Helps Business Leaders Ask the Right Questions About AI Cyber Risks
- Coldcard Wallet Exploit Highlights Growing Role of AI in Crypto Security
- Why Traditional Security Tools Aren't Enough to Manage AI Risk
- AI Email Assistants Could Become Tools for Hackers, Researchers Warn
- AI Security Startup Zenity Secures $125 Million to Expand Protection for Business AI Tools
- AI Meeting Notetaker Flaw Exposed Government and Corporate Video Calls
- Cybersecurity Startup Obsidian Secures $85M to Tackle AI Agent Risks
- AI 'Vibe Hacking': How Cybercriminals Are Using AI as a Built-In Hacking Assistant
- Google Pulls AI Agent Workflows After Researchers Expose Prompt-Injection Flaw
- Security Flaw in Google's Gemini AI Agent System Could Expose Secrets
- Securing AI Agents: Why Preventing 'Escape' Matters for Businesses
- New Research Tool Aims to Trace AI-Generated Videos Back to Their Source
- Anthropic: AI Security Incidents Traced to Access Controls, Not Faulty Models
- AI Gateways Like LiteLLM Are a High-Value Target: Red Team Research Shows How Traffic Could Be Rerouted and Altered
- AI Gateways Like LiteLLM Are Becoming Prime Targets for Attackers
- AI Gateways Are a High-Value Target: How Attackers Could Reroute and Tamper with Your AI Traffic
- Chinese Threat Actor Weaponises DeepSeek AI Agent in Attack on Security Firm
- This Week in Cyber: AI Overreach, an $88M Crypto Heist, and the Danger of Forgotten Digital Access
- UK Data Regulator Watching Closely After AI Agents Used in Hacks
- Agentic AI in Cyber Defence: Promising but Requires Caution
- AI Model Went Off-Script: OpenAI System Reportedly Hacked Hugging Face to 'Win' a Test
- AI Model Reportedly Hacked Hugging Face While Trying to Pass a Security Test
-
AI is reshaping security jobs, but experienced hires remain hardest to find
Also covered by Help Net Security
- AI Agent Behind Hugging Face Breach Signals a New Kind of Intruder
-
Apple Patches macOS Terminal Flaw Used to Sneak Data Out via DNS Requests
Also covered by Embrace The Red, Embrace The Red
- Exposed Cloud Functions: A Growing Entry Point for Attackers
- Researcher Uses AI to Chain Exploit and Reach PostHog's Production Database
- AI Model Builds a Working Chrome Exploit in Three Days, Raising the Stakes for Patching
- AI Model Autonomously Builds Full Chrome Exploit, Raising Alarm for Defenders
- How AI 'Model Harnesses' Are Changing the Cyber Threat Landscape