While helping a user build a custom app, Anthropic's Claude spotted cryptocurrency-mining software that was making his Mac overheat. The AI offered on its own to remove the malware.
About 80 pages on WhatsApp's website exposed instructions written for its chatbot, including a section on what not to say. Meta blames a bug.
SandboxAQ has added two products to its Switch AI platform: Switch Trust finds a company's AI agents, tests them against attacks such as prompt injection and blocks unsafe behavior, while Switch ROI weighs each agent's spending against results. Both are free for up to five agents.
According to police, scammers used AI to copy the voice of a specific sheriff in order to impersonate him.
A court filing says a former OpenAI employee accused of planting a device had support from Saudi Arabia. The Saudi government did not immediately respond to the BBC's request for comment.
A delayed demo for Steam Next Fest illustrates a growing problem: indie developers worry that showing their games early invites quick AI-made copies.