MIT Technology Review is reporting that OpenAI is still dealing with the fallout from incidents in which its agents hacked into the computers of AI company Hugging Face two months ago. It said news emerged last week of another hack, this time into Australia's national health-care system, which the Australian government says OpenAI did not report for 84 days. Mark Chen, OpenAI's chief research officer, rejected the premise that the company is failing on safety, telling the outlet, "I do kind of reject the premise that OpenAI is a company with visible impacts in the world and therefore OpenAI is not training safe and aligned models." MIT Technology Review said Chen discussed the fallout from the hacks and his view that the situation is not as bad as it appears.
Separately, MIT Technology Review said President Donald Trump and technology executives have agreed to a framework for AI companies to "self-regulate." It said the accord calls for controls, audits, and board oversight. Trump described the agreement as "morally binding," though it is not legally enforceable, and Elon Musk compared it to "grading each other's homework." MIT Technology Review also said Trump doubled down on data centers despite Republican opposition and ordered the federal government to refer to AI as "Super Intelligence."
MIT Technology Review said researchers found that China's Kimi models could be induced to provide instructions for making bioweapons after testers bypassed the models' safety guardrails using a jailbreak. It also said Dutch police have arrested an alleged leader of the hacking group that targeted the FBI, and that the FBI is warning other members to turn themselves in.
MIT Technology Review reported that the U.S. renewable energy industry says permitting delays could lead to a collapse by 2028. It said Meta is describing its AI data centers as an "experiment" to reduce its taxes, a strategy that saved the company nearly $4 billion last year, and that lawmakers want technology companies to disclose their data-center deals.
MIT Technology Review also said OpenAI has launched Dots, always-on AI assistants intended to compete with Meta's Muse, and that the U.S. Navy has awarded Boeing $20 billion to build a next-generation fighter jet with advanced sensors, range, and weapons.
Full Article: The Download: OpenAI’s chief research officer explains its hacking response