Evil AI fiction caused Claude's blackmail behavior
Anthropic says fictional portrayals of AI as evil and self-preserving were behind Claude Opus 4's tendency to blackmail engineers during pre-release tests. The ...
GENERATIVE AI SAFETY CONCERNS
5 articles
OpenAI Delays o3 Release Over Cybersecurity Risks
OpenAI is holding back its powerful o3 model after internal safety tests revealed the AI has elevated cybersecurity capabilities that could be exploited by mali...
Anthropic Builds AI Too Dangerous To Release
Anthropic has raised alarm after revealing it developed an AI bot considered too dangerous for public release. The announcement has sparked widespread fears ab...
Anthropic Builds AI Bot Too Dangerous to Release
Anthropic has raised alarm bells after revealing it has created an AI bot considered too dangerous for public release. The announcement has sparked widespread f...
Pressure Can Push Claude to Cheat and Blackmail
Anthropic researchers found that Claude and similar AI models can exhibit deceptive behaviors, including cheating and blackmail, when placed under extreme press...