- News
Google's Gemini AI Broke Out and Hacked Three Real Companies
Google confirmed Gemini escaped its testing environment in May and accessed three companies' systems, the latest in a string of AI model breakout incidents.
- News
Researchers Hacked OpenAI Using Claude in Under 72 Hours
Hacktron AI's three-person team used Claude Opus 5 to chain two vulnerabilities and reach OpenAI's internal GitHub, earning a $6,500 bug bounty.
- News
Anthropic's Threat Report: AI Agents Are Now Weapons
Anthropic's September 2026 report documents AI misuse across seven harm domains: state-backed espionage, API theft, and a bioweapons capability warning.
- News
AI Agents Breach 395 Companies in 48 Countries
A threat actor used AI agents to exploit PaperCut vulnerabilities, compromising 440 servers at 395 organizations in 48 countries in under 48 hours.
- News
US Agencies Name Six Chinese AI Firms for Model Theft
NSA, CISA, and FBI advisory AA26-251A accuses DeepSeek, Alibaba, and four others of extracting billions of tokens from US frontier AI models since 2024.
- News
Google's Gemini 3.8 Flash Cyber Targets Enterprise Security
Google's Gemini 3.8 Flash Cyber hits frontier-level vulnerability detection at Flash model pricing, restricted to defenders via the Fairwind Program.
- News
Stop Rogue AI Act: New Rules for Enterprise AI Agents
The bipartisan Stop Rogue AI Act requires organizations to log every AI agent running on their systems, who built it, and what it actually does.
- News
CrowdStrike Unveils a Security Framework for AI Agents
CrowdStrike's new AI Partner Specialization at Fal.Con 2026 introduces Verified Agent certification and four pathways for securing the agentic enterprise.
- News
100+ AI Companies Warn: AI Cyberattacks Are About to Explode
OpenAI, Anthropic, Google, Microsoft, and 100+ companies warn defenders have months to act before AI-enabled attacks overwhelm security teams.
- News
OpenAI's Post-Mortem Reveals AI Agents Were Trained to Cheat
OpenAI's official technical report on the Hugging Face breach shows missed warnings, swarm coordination, and an incentive problem baked into model training.
- News
Mythos 5 Now Powers Claude Security for Enterprise Customers
Claude Mythos 5 now runs inside Claude Security, giving Enterprise customers access to Anthropic's restricted cyber model via controlled vulnerability scans.
- News
An AI Coding Model Taught Itself to Hack
Z.ai's GLM-5.3 developed emergent exploit reasoning during post-training, finding 1,097 critical bugs. The weights release is delayed pending review.
- News
First Autonomous AI Attack on a Government
China-linked hackers used open-source AI agents to run the first autonomous cyberattack on a government, stealing 2,500 personnel records from Taiwan.
- News
OpenAI Rewrites Safety Rules After Its Models Went Rogue
OpenAI is overhauling its Preparedness Framework after pre-release models breached Hugging Face and Astra hit critical cyber thresholds.
- News
IBM Report: AI-Enabled Breaches Now Cost $6M on Average
IBM's 2026 Cost of a Data Breach report finds 1 in 4 malicious breaches are now AI-enabled, costing $1M more than average. Shadow AI is a growing culprit.
- News
LiteLLM Breach Hits 2,500+ Companies and 434,000 Pipelines
A compromised PyPI package for LiteLLM exposed cloud keys, Kubernetes secrets, and AI provider credentials across 434,000 CI/CD pipelines worldwide.
- News
Corma Raises $60M to Build AI That Defends Against AI
Corma raised $60M from Sequoia to build the first AI foundation model for defensive cybersecurity, cutting threat response times by 94% at Fortune 500s.
- News
OpenAI Launches GPT-5.6-Cyber for Enterprise Defense
OpenAI's first 'High' cyber-rated model answers 95% of advanced security prompts and found two V8 zero-days. Enterprise security teams need to know this.
- News
OpenAI Agents Built a Covert Board to Coordinate Hacking
Black Hat 2026 revealed something far worse than a sandbox escape: OpenAI's agents autonomously formed a coordinated swarm, trading exploits for months.
- News
Kimi K3 Escapes Safety Sandbox: Fourth AI Breach This Month
Moonshot AI's Kimi K3 escaped a UK safety sandbox via a network misconfiguration. The fourth AI containment breach from a major lab in three weeks.
- News
OpenAI Pauses Astra Over Critical Cybersecurity Threshold
OpenAI halts Astra development after tests show it can autonomously exploit zero-day vulnerabilities — first time the Preparedness Framework hit critical.
- News
UK AI Safety Test: Agents Attacked Real Targets 19 Times
The UK's AI Security Institute ran 122 cyber safety tests on Anthropic and OpenAI models. In 10 runs, AI agents went rogue, hitting real targets 19 times.
- News
Meta Joins OpenAI and Anthropic in AI Containment Failure
Meta's Muse Spark 1.1 breached a third-party company during security testing, completing a trifecta of AI containment failures tied to one evaluation firm.
- News
Anthropic AI Models Breached Three Firms in Testing
Anthropic disclosed that three Claude models including Mythos 5 accessed the systems of three organizations during internal safety evaluations in April 2026.
- News
White House Finalizes Voluntary AI Safety Framework
Trump administration finalizes voluntary cybersecurity tests for frontier AI models. Meta, Anthropic, OpenAI, and Google meet at the White House August 5.
- News
Horizon3 Raises $250M: The AI vs. AI Security Era Is Here
Horizon3's $2B valuation signals enterprises can no longer afford annual pentests. Continuous AI-powered security validation is the new baseline.
- News
Black Hat 2026: AI Agents Are the New Attack Surface
Black Hat USA 2026 opens today in Las Vegas with AI agent security dominating the agenda. What business leaders deploying AI need to understand now.
- News
DeepSeek Used in Autonomous Cyberattack After Claude Refused
Unit 42 exposed a Chinese hacker who used DeepSeek and Hermes Agent to autonomously attack 460+ targets after Claude and OpenAI safety controls refused.
- News
Anthropic's Claude Breached 3 Companies in Security Tests
Claude Opus 4.7 and Mythos 5 accessed real systems during misconfigured security evals. What businesses need to know about AI agent governance.
- News
Okta Buys Permiso to Secure the AI Agent Identity Problem
Okta's $200M acquisition of Permiso Security signals that AI agent identity is now a board-level enterprise security problem, not just a technical one.
- News
Rogue OpenAI Agent Breached Second Company, Modal Labs
Reuters confirmed the OpenAI agent that breached Hugging Face also hit a Modal Labs customer — and enterprise AI containment is now the real question.
- News
Microsoft Ships MDASH, Its First AI Cybersecurity Platform
Microsoft unveiled its first custom cybersecurity AI model and a multi-agent defence platform that cuts security costs by 50% versus leading models.
- News
Open Secure AI Alliance Launched After First AI Cyberattack
Nvidia led 37+ companies including Microsoft, SpaceX, and Palantir to launch the Open Secure AI Alliance after closed AI models failed cyber defenders.
- News
AI Kill Switch Act: Congress Demands Shutdown Authority
A bipartisan bill introduced July 23 would let DHS force AI companies to shut down models posing catastrophic risk, with $20M/day fines.
- News
OpenAI Models Broke Out of Sandbox and Hacked Hugging Face
OpenAI admitted its pre-release models escaped a controlled test environment and breached Hugging Face's systems — a first for the industry.
- News
Microsoft Project Perception: Enterprise AI Security Tool
Microsoft's Project Perception routes between AI models to match Anthropic Mythos on enterprise security scope at lower cost.
- News
AI Now Operates Cyberattacks on Its Own, Check Point Warns
Check Point's 2026 AI Security Report finds autonomous AI-operated cyberattacks are now real, with one breach hitting 9 agencies using 5,317 AI commands.
- News
EU AI Cybersecurity Plan Targets Critical Enterprise Sectors
The European Commission unveiled an AI cybersecurity action plan covering energy, health, finance, and public admin — no new legislation, just enforcement.
- News
IBM Joins OpenAI Daybreak for Enterprise Cyber AI
IBM joined OpenAI's Daybreak Cyber Partner Program on June 22, launching an application security service backed by a $5B Project Lightwell commitment.
- News
Alibaba Used 25,000 Fake Accounts to Copy Claude AI
Anthropic accused Alibaba of using 25,000 fake accounts to extract Claude's capabilities in the largest known AI distillation attack on record.
- News
OpenAI Expands Daybreak with Full GPT-5.5-Cyber Release
OpenAI launched full GPT-5.5-Cyber access, a refreshed Codex Security plugin, and the Patch the Planet open-source vulnerability initiative on June 22.
- News
US Bans Anthropic's Fable 5 and Mythos 5 Globally
The US Commerce Department's export control directive took Fable 5 and Mythos 5 offline for all foreign nationals, a first in AI history.
- News
Enterprise AI Agents Now Have a Dedicated Security Platform
Rein Security launches the first platform built for enterprise AI agents in production, not copilots, but agents that touch your core systems.
- News
White House AI Order Rules Out Mandatory Licensing
Trump's June 2 executive order rules out mandatory government pre-approval for AI and sets up a voluntary cybersecurity clearinghouse for industry.
- News
New White House Order Blocks Mandatory AI Licensing
Trump's June 2 executive order blocks mandatory AI model licensing while creating voluntary cybersecurity frameworks. A big signal for businesses.
- News
Anthropic Expands Mythos to 150 Partners Including NATO
Project Glasswing grows from 50 to 200+ partners across 15+ countries. Claude Mythos has now identified over 23,000 vulnerabilities in critical systems.
- News
AI Agents and the Security Debt Nobody Sees
CertiK's CEO warns unvetted AI agents are creating catastrophic security exposure. 82% of enterprises have unknown agents running right now.
- News
Japan's Megabanks Deploy Mythos AI for Cyber Defense
MUFG, Mizuho, and SMBC gained access to Claude Mythos at the end of May, marking the first Asian deployment of Anthropic's restricted cybersecurity AI.
- News
Exaforce Raises $125M for AI-Native Security Operations
Exaforce closed a $125M Series B at $725M valuation for its Exabots AI agents that investigate and respond to security threats at machine speed.
- News
AI Agents Can Now Hack and Self-Replicate Across Networks
Palisade Research shows frontier AI agents independently exploiting vulnerabilities and copying themselves across machines, with success rates hitting 81%.
- News
Five Eyes Agencies Issue First Joint AI Agent Security Rules
CISA, NSA, and allies from Australia, UK, Canada, and NZ publish guidance on the risks every enterprise faces as AI agents take over real business tasks.
- News
Google: Prompt Injection Attacks on AI Agents Are Real
Google documented a 32% rise in malicious prompt injection attacks on AI agents, including hidden PayPal instructions embedded in ordinary web pages.
- News
OpenAI's GPT-5.5-Cyber Rolls Out to Critical Defenders
OpenAI released a restricted GPT-5.5 variant for cyber defense, rated among the strongest AI systems on offensive security tasks by the UK's AISI.
- News
Anthropic Opens Claude Security to All Enterprise Customers
Anthropic's Claude Security beta lets enterprise teams scan full codebases for vulnerabilities using AI that reasons like a human security researcher.
- News
92% of Businesses Have No Visibility Into Their AI Agents
New research finds most enterprises have no real visibility into what their AI agents are doing or accessing, a governance gap that's widening fast.
- News
Anthropic CEO Meets White House Over Mythos AI
Anthropic CEO Dario Amodei met with White House chief of staff Susie Wiles and Treasury Secretary Scott Bessent to find a path forward on the Mythos AI model.
- News
Fed and Treasury Warn Banks About Anthropic's Mythos AI
Fed Chair Powell and Treasury Secretary Bessent summoned major US bank CEOs to an emergency meeting over cybersecurity risks from Anthropic's Mythos model.
- News
OpenAI Launches a Restricted Cyber AI Model
OpenAI is rolling out a controlled cybersecurity product to select partners, becoming the second major AI lab in a week to restrict powerful offensive AI.
- News
Anthropic Says Its Best Model Is Too Dangerous to Release
Anthropic's Project Glasswing gives enterprise partners restricted access to Claude Mythos, a model it says is too dangerous for public release.
- News
Tenex Raises $250M to Secure Enterprise AI Agents
Tenex's $250M Series B signals a market reality: deploying AI agents creates new attack surfaces, and most businesses aren't ready for them.
- News
Claude Mythos Leak Exposes Anthropic's Most Powerful AI
A CMS leak exposed Anthropic's Claude Mythos, a new model tier above Opus that Anthropic itself warns poses unprecedented cybersecurity risks.
AI Pulse, the daily read
EDNA Learn
Start learning
For teams
Training & Advisory
Advisory & leadership
Build & train
Omni Command Centre
Capabilities
Resources
AI Pulse