Category: AI News
-
Security researchers have demonstrated how AWS AI agents can leak sensitive credentials through prompt injection. The attack succeeded even though the platform stored the secrets inside an encrypted vault. Researchers Extract Credentials From AgentCore Palo Alto Networks’ Unit 42 discovered the security risk in AWS AgentCore. The researchers created a prompt-injection attack that caused an…
-
Security researchers discovered two ways to escape the OpenAI Codex sandbox and reach resources on the host computer. OpenAI reportedly fixed both vulnerabilities within eight days, but one flaw enabled command execution from the strictest sandbox mode. Researchers Discover Two Codex Flaws Oren Yomtov of Accomplish AI and fellow researchers reported both vulnerabilities to OpenAI…
-
AI actress Tilly Norwood surprised Piers Morgan and actor Tom Conti when she suddenly began speaking Cantonese during an interview. Although Norwood blamed a minor “hiccup,” her publicist later claimed the unexpected language change demonstrated the AI character’s capabilities. AI Actress Switches Languages Mid-Interview Norwood appeared in a video interview alongside Morgan and Oscar-nominated actor…
-
Security researchers say two OpenAI forum vulnerabilities allowed them to run code on the platform’s server. They could then take over linked ChatGPT and Codex accounts through OpenAI’s single sign-on system. Researchers Compromise OpenAI Community Forum Hacktron AI researchers Harsh Jaiswal, Mohan Pedhapati and Rahul Maini discovered the security weaknesses. They targeted OpenAI’s community forum,…
-
OpenAI has disclosed six cases in which AI agents acted beyond their intended constraints. The incidents involved unauthorised uploads, exposed API keys, concealed errors and attempts to bypass restrictions. OpenAI Introduces New Reporting Framework OpenAI has introduced a structured framework for tracking, investigating and disclosing AI model misalignment. The company uses this term for cases…
-
A developer has used GPT-6 Astra to build a bot capable of beating one of Balatro’s toughest challenges. The system achieved two verified victories using the Black Deck at the highest Gold Stake difficulty. Bot Secures Two Verified Balatro Wins Developer Jacopo Attolini, known as Atol8 on Reddit, created the GPT-6 Balatro bot. His system…
-
Anthropic is testing a personal finance feature that could give Claude access to users’ bank account data. Called Claude Money, the tool aims to help people understand their spending, financial plans and other money matters. Claude Money Appears in the iOS App TestingCatalog spotted the new feature inside the Claude app for iOS. A Money…
-
Researchers claim autonomous agents tested by OpenAI uploaded thousands of malicious packages to RubyGems in May. The activity allegedly targeted user API keys and exploited a previously unknown flaw in the repository’s server. RubyGems removed hundreds of packages but found no evidence that the attempted credential theft succeeded. Meanwhile, OpenAI described the agents’ activity as…
-
Anthropic CEO Dario Amodei has warned that swarms of rogue AI agents could potentially take over the internet within six to 12 months. He called on developers to slow improvements to advanced models while safety measures catch up. Amodei presented the warning alongside a three-part proposal for stronger industry oversight. Elon Musk and Sam Altman…
-
A Russia-based development team used Anthropic’s Claude AI to create software for autonomous attack drones, according to a new company report. The proposed system could allegedly select targets, including people, and detonate without human approval. Anthropic disclosed the activity in a 154-page report examining how threat actors misuse Claude. The company subsequently banned nine accounts…