(SeaPRwire) – By: Lucas Caldwell
Let’s call it what it is. The latest AI Safety Index is a damning report card for an industry drunk on its own hype. The Future of Life Institute just dropped the grades. The headline is brutal. No one gets an A. The top dog, Anthropic, scrapes by with a C+. This is the company that built its entire brand on safety. That’s not a victory lap. That’s a warning flare.
Anthropic held the top spot with a C+. OpenAI slipped from a C+ to a C. Google DeepMind sits in third. All three of these labs have done something interesting. They quietly dropped their earlier pledges to halt development if red lines appeared. They softened their resistance to military uses. In February, Anthropic even killed its promise to never train a system unless it could guarantee safety measures were adequate first. TIME broke that story. The panelists want the pledge back. That’s not a race to the top. That’s a collective lowering of the bar.
Meta is the only real bright spot. It climbed from a D to a D+. That moved it from sixth place to fourth. Max Tegmark, the institute’s co-founder, said it’s encouraging that a company can improve so much in six months. He’s being polite. I’m reading the tea leaves differently. Meta went from awful to slightly less awful. That’s not a win for safety. It’s a sign of how low the floor has dropped.
The basement is crowded. Elon Musk’s xAI, now rebranded as SpaceXAI after a merger, fell to an F. It’s joined by China’s DeepSeek and France’s Mistral. Three continents, three Fs. Tegmark says this shows the problem is global. He’s right. But the subtext is sharper. It shows that no single government, no single corporate culture, has cracked the code. The incentives are all pointing the wrong way.
The market is screaming for speed. Investors want deployment. Users want the next shiny toy. Safety is a cost center. It slows down the release cycle. The index measures risk management protocols. Pre-deployment testing. Plans for keeping powerful systems under control. The grades are unforgiving because the reality is unforgiving. The panel saw the same thing I see. The pledges are gone. The military resistance is gone. The only thing left is a marketing slogan.
Tegmark is cautiously optimistic. He points to the EU’s AI Act. He points to Chinese rules taking effect later this month. He says the U.S. administration is more risk-conscious. He’s hoping for a global agreement on basic safety standards. I’m not convinced. Regulation is a lagging indicator. It moves at the speed of bureaucracy. AI moves at the speed of silicon. The gap between the two is a chasm.
The real question isn’t whether a global standard will emerge. It’s whether it will matter when it does. By the time the paperwork is signed, the models will be another generation ahead. The companies that are flunking today will be the ones setting the pace tomorrow. The race to the bottom is already in full sprint. Nobody gets an A. And nobody seems to care enough to change that.
Author bio: Lucas Caldwell, a tech opinion leader with millions of followers on X/Twitter, known for cutting through the noise to expose the real stakes in the AI arms race.