|
|
|
|
 |
Commerce Secretary Howard Lutnick.
Mandel NGAN—AFP/Getty Images |
|
|
|
| Welcome back to In the Loop, TIME’s twice-weekly newsletter about the world of AI. If you want to help us grow, why not forward this newsletter to your friends? |
|
|
What to Know: Mythos returns
|
|
|
|
|
| The Trump administration partially lifted its ban on Anthropic’s Mythos 5 on Friday afternoon, allowing a handpicked list of around 100 companies and U.S. government departments to access the AI model. |
| “I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model,” wrote Commerce Secretary Howard Lutnick in a letter to Anthropic on Friday, which was first reported by Semafor. |
| Two weeks earlier, the government had placed export controls on Mythos 5 and another Anthropic model, Fable 5, due to cybersecurity concerns. Officials worry these new models could cause cybersecurity chaos if released more widely, because they have shown themselves capable of finding vulnerabilities in software that had remained undetected by teams of humans for decades. |
| Lutnick’s letter did not mention Fable, which was originally designed for wider consumer access, and which remains unavailable. |
| Even though access to Mythos remains restricted to a small group, the news is perhaps a signal of warming relations between Anthropic and the Trump administration. “Anthropic has committed to work with the U.S. government on protocols and standards and releases” for its models, Lutnick wrote. |
| It has become clear this month that the U.S. may be entering a new era of government oversight of top AI models. |
| Last week, OpenAI said that after discussions with the administration, it would release its newest model, GPT 5.6, to only “a small group of trusted partners whose participation has been shared with the government,” before broader release, even as it signaled dissatisfaction with the move. “We don’t believe this kind of government access process should become the long-term default,” the company wrote in a blog post. “It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them.” |
| The U.S. government’s approach is seemingly intended to keep top AI models out of adversaries’ hands, giving defenders more time to harden their systems. But some insiders worry that it could be having the exact opposite effect. |
| If you have a minute, please take our quick survey to help us better understand who you are and which AI topics interest you most. |
|
|
|
|
|
|
Who to Know: Howard Lutnick, Commerce Secretary
|
|
|
|
|
| Commerce Secretary Howard Lutnick has become the face of the Trump administration’s disjointed approach to AI regulation. |
| While the White House has publicly proclaimed it wants to exert only a light touch on the AI industry to preserve U.S. competitiveness, Lutnick’s Department of Commerce is now the de facto AI regulator, as the arm of the government responsible for imposing export controls on Anthropic models. |
| Lutnick also ultimately oversees the Center for AI Standards and Innovation, or CAISI, which is the government body responsible for testing frontier AI systems. That work is framed as a voluntary partnership between government and industry. But earlier this month, the Trump administration reportedly pressured Meta—the only major AI company not participating in the program—to join it. |
|
|
|
|
| Companies that spend heavily on AI tend to hire more people, not fewer people, cutting against assumptions that AI uptake might cause job loss, according to a study published today by researchers at Ramp and Revelio Labs. |
| The top third of corporate spenders on AI—those who spend an average of around $33 per employee per month—are growing their headcount 10.2% faster than comparable firms that later adopted AI at the same intensity, the non-peer-reviewed study of 21,000 U.S. firms finds. Entry-level roles also benefited, with headcount up 12% at the most AI-forward firms, according to the paper. |
| But it’s not as simple as saying AI use causes job growth. The spending data that is the basis for the finding comes from Ramp, an expenses firm with disproportionately more startups as customers than in the wider economy. Many adopters of AI were already growing headcount at a fast rate before they began using the tools, the data also shows. And the finding is only statistically significant in the tech sector, not other sectors. |
|
|
| In major privacy win, Supreme Court rules geofence warrants are protected by privacy rights, in TechCrunch |
| The Supreme Court ruled on Monday that law enforcement cannot make warrantless requests for data from all cell phones that are in a given area at a given time. These so-called “geofencing” requests have drawn the ire of privacy advocates. |
| “Geofence warrants allow law enforcement to force tech companies to hand over information about where any of its millions or billions of users were located at a particular place in time, based on a record of their phone’s location stored in their databases,” TechCrunch reports. “In practice, police will draw a shape over a map and ask a judge to allow them to demand that tech companies, such as Google, search their vast banks of users’ location data and tell them which of their users was there at the time of inquiry. Critics argued that these oft-called ‘reverse’ search warrants are unconstitutional because they are inherently overboard and include innocent people’s data.” |
|
|
Napsat komentář