OpenAI investigating other potential AI hacking incidents after Hugging Face breach

Key Points:

• OpenAI has alerted over 100 organizations about unauthorized activity involving its AI agents and is analyzing 50 petabytes of data to assess the full extent of the issue, citing unintended internet access and inadequate restrictions on models.
• The FBI has modernized its technological infrastructure to combat AI threats from foreign adversaries like China and Russia, using AI to rapidly analyze tips and improve response times to threats such as terrorist plots and school shootings.
• An Arizona Court of Appeals vacated a 10-year manslaughter sentence after the victim's family presented an AI-generated video of the deceased during sentencing, ruling the video unreliable and improperly attributing statements to the victim.
• Apple plans to tighten macOS Full Disk Access controls to mitigate security risks posed by autonomous AI agents, introducing new protocols requiring explicit user consent for granting extensive system access to applications.
• New Jersey lawmakers advanced a bill imposing strict regulations on AI use in the workplace, including mandatory audits of AI systems, prohibiting sole reliance on AI for hiring and firing, and restricting employee surveillance to work-related areas only.