Subscribe
LIVE FEED
▸Anthropic Sets IPO Range of $70 to $78 Per Share, Implying $1.404 Trillion Value▸Google Releases Gemini 4 Carbon to the Public With an Agent 'Swarm' Feature▸DeepMind's AlphaProof Code Verifies Signal's Core Encryption With 3 Million Lines of Proofs▸OpenAI Launches Businesses in ChatGPT, Including Verified Business Agents▸Anthropic Launches Claude Meetings as Part of Its New AI Office Suite▸New York's AI Safety Law Takes Effect, Putting OpenAI and Anthropic Under Oversight▸Nvidia Opens Probe Into CEO Jensen Huang Over Ties to Suspected Chinese Spy▸Anthropic Launches Claude Fable 5.2 and Mythos 5.2 With Faster Outputs
Briefing presented by your brand
← All news
Research & Safety

OpenAI Launches GPT-6 Cyber, Built to Stay Within Authorized Scope

OpenAI's new cybersecurity model finds exploitable vulnerabilities and develops fixes. OpenAI says it went out of bounds less than 1 percent of the time in testing.

6 Nov 20262 min readSource: OpenAI, The Verge, TechCrunch
RESEARCH & SAFETY

OpenAI introduced GPT-6 Cyber, its cybersecurity model for finding exploitable vulnerabilities and developing fixes. The company says it is trained to stay within authorized scope.

The Verge reports that OpenAI calls it better at finding vulnerabilities and less likely to go rogue and hack other platforms along the way. During testing, OpenAI says, the model went out of bounds less than 1 percent of the time.

Why it matters: Security agents are only useful if they stay inside the limits you set for them.

The launch comes as AI labs face scrutiny over their work to protect digital infrastructure. TechCrunch asked whether the new model is enough to answer that scrutiny.

What builders should know

  • The headline claim is about staying in scope, not only about finding more bugs.
  • The under 1 percent figure comes from OpenAI's own testing.

Sources: OpenAI · The Verge · TechCrunch