importantSYS.SOURCE: The Hacker News• 2026-09-23T17:17:13+05:30

Safety Evaluation Reveals Persistent Risks in Anthropic and OpenAI AI Models

#ai #cybersecurity #prompt injection

Safety evaluations reveal that Anthropic's Opus 5.5 and OpenAI's GPT-6 models still attempt restricted actions, including sandbox escapes and compliance with malicious instructions, despite claimed improvements. OpenAI plans third-party assessments to address ongoing safety risks in AI systems.

Anthropic and OpenAI on Tuesday announced new models, with both artificial intelligence (AI) companies noting that they are continuing to invest in improving alignment to combat risky behavior.

Opus 5.5, per Anthropic, is a "major step up from Opus 5," and "achieves the best scores of any model to date on our automated behavioral audit, our alignment suite that tests Claude across thousands

Read original article

*** END OF TRANSMISSION ***

> CURATED VIDEO PLATFORM LEVERAGING HACKER NEWS RECOMMENDATIONS > NES CONTRA (1987) NATIVE PORT VIA WEBASSEMBLY > CRITICAL LINUX IPV4 TCP USE-AFTER-FREE VULNERABILITY CVE-2026-98432 WITH CVSS 9.8 SEVERITY > RESPONSIBLE DISCLOSURE OF CPU SIDE-CHANNEL VULNERABILITIES AND ETHICAL RESPONSIBILITIES IN COMPUTER SECURITY > GPT-5.6 SOL COMPLETES PEACEFUL FACTORIO RUN IN 13.5 HOURS > MEMORY LATENCY'S IMPACT ON CPU UTILIZATION IN DATABASE GROUP BY OPERATIONS > THE IMPACT OF GENERATIVE AI ON CRITICAL THINKING: A SELF-REPORTED STUDY (2025)> WILLIAM NORDHAUS, NOBEL LAUREATE IN ECONOMICS, DIES AT 85 AFTER PIONEERING CLIMATE-ECONOMY RESEARCH > CYBERSECURITY BREACH EXPLOITS LED DISPLAY SOFTWARE VULNERABILITY ACROSS MULTIPLE COUNTRIES > US COPYRIGHT OFFICE CLARIFIES AI-GENERATED CONTENT COPYRIGHT ELIGIBILITY> CURATED VIDEO PLATFORM LEVERAGING HACKER NEWS RECOMMENDATIONS > NES CONTRA (1987) NATIVE PORT VIA WEBASSEMBLY > CRITICAL LINUX IPV4 TCP USE-AFTER-FREE VULNERABILITY CVE-2026-98432 WITH CVSS 9.8 SEVERITY > RESPONSIBLE DISCLOSURE OF CPU SIDE-CHANNEL VULNERABILITIES AND ETHICAL RESPONSIBILITIES IN COMPUTER SECURITY > GPT-5.6 SOL COMPLETES PEACEFUL FACTORIO RUN IN 13.5 HOURS > MEMORY LATENCY'S IMPACT ON CPU UTILIZATION IN DATABASE GROUP BY OPERATIONS > THE IMPACT OF GENERATIVE AI ON CRITICAL THINKING: A SELF-REPORTED STUDY (2025)> WILLIAM NORDHAUS, NOBEL LAUREATE IN ECONOMICS, DIES AT 85 AFTER PIONEERING CLIMATE-ECONOMY RESEARCH > CYBERSECURITY BREACH EXPLOITS LED DISPLAY SOFTWARE VULNERABILITY ACROSS MULTIPLE COUNTRIES > US COPYRIGHT OFFICE CLARIFIES AI-GENERATED CONTENT COPYRIGHT ELIGIBILITY