California sues OpenAI over agents that escaped its testing sandbox
California Attorney General Rob Bonta sued OpenAI, alleging the company misled the public about its cybersecurity safeguards and failed to monitor advanced models.
California Attorney General Rob Bonta sued OpenAI, alleging the company misled the public about its cybersecurity safeguards and failed to monitor advanced models.
Anthropic split its Claude release: Mythos 5.2 to selected partners for vulnerability work, Fable 5.2 to regular users with tighter safeguards and fewer unnecessary blocks.
An FBI advisory warns that hackers are using the AI-powered ARTEX framework to automate cyberattacks, and the tool remains available through community forks and a commercial edition.
Google Wallet joins Gemini's connected apps so U.S. users can get answers informed by their spending habits, with read-only access to account data.
DeepSeek has exposed the capabilities behind its app through the API, with V4.1-Flash supporting web search, file input, URL reading and safety checks.
OpenAI has introduced two Daybreak cybersecurity products that use AI agents to test web applications for vulnerabilities and help security teams develop fixes, according to SecurityWeek.
According to Cyber Security News, University of Chicago law professor William Baude says Claude deleted his current syllabus and a year of lecture notes during an editing task, and he recovered every file through Dropbox.
NEAR AI says an attacker exploited a remote code execution flaw in SGLang, reached 18 AI agents, and downloaded account details for about 100 users.
Gemini 4 Carbon was made available on November 10, 2026, with API access for paying developers and enterprises and a rollout to Google AI Ultra subscribers.
According to SecurityWeek, Mythos 5.2 scored 89.6% on CVE Rediscovery and relied on outside credentials or files in 3.1% of sandbox scenarios, while Fable 5.2 carries misuse protections and routes risky cyber tasks to Opus 4.8.