Author: Manuel Bosque

A new enterprise AI report says some of the biggest corporate buyers, including Nvidia, Palantir, and Booz Allen Hamilton, are scaling back or flat-out stopping their use of public OpenAI and Anthropic models because they don’t trust where the data might end up.

According to the report, the worry doesn’t stop at training data. These companies are also looking hard at prompts, responses, logs, usage patterns, and metadata, since any of that can reveal connected apps or expose how internal processes actually work.

OpenAI and Anthropic both say their enterprise contracts don’t train on customer data by default. But C Spire and others argue that technical data can still be collected, without clear guardrails around what gets captured, how long it’s stored, or who can access it. The report says Anthropic’s June policy change made those concerns sharper by allowing some retention for abuse monitoring and, in research cases, for up to 30 days.

For companies that handle regulated data or valuable IP, the direction is hard to miss. The report says zero-data-retention terms, air-gapped environments, private cloud, and on-premises deployments are gaining traction. It also says nearly three-quarters of S&P 500 companies now treat generative AI as a material risk, while a 2024 study found that 27% had temporarily banned it.

Companies can still use public models through those providers. Even so, the report argues that isolated deployments may be worth the higher upfront cost, because at enough scale they can beat per-token pricing.

Transformative AI Strategy for Europe is out, and it opens with a pretty blunt warning. Its authors say Europe needs to raise its share of advanced-model compute from about 5% to 15% by 2030. If it doesn’t, they argue, the region risks becoming a “playground for foreign powers.”

Philippe Aghion, Margrethe Vestager, Monika Schnitzer, and Marta Kwiatkowska write that the European Union held only about 5% of global compute capacity in 2026, compared with 78% for the US and 11% for China. The point they’re making is simple: if Europe wants to build, train, and deploy AI systems that can actually compete, regulation by itself won’t get it there.

From there, the paper argues for a major build-out of data centers, tied to clear local benefits, plus easier scaling for startups and a smarter use of chokepoints such as ASML to improve access to frontier compute. It also notes that AI investment reached 6.8 billion in 2025, up 83.3%, while enterprise adoption averaged 20% across the European Union. That figure ranged from 42% in Denmark to 5.2% in Romania, and smaller firms are still trailing.

And compute isn’t the only issue on the list. The paper also calls for more frontier technical talent inside European Union institutions, more technical enforcement of systemic-risk rules, stronger cyber protections, crisis stockpiles for biological and other catastrophic risks, and European leadership in assurance technology.

You can read the paper online.

Anthropic published We Must Pace the Frontier in September 2026. The proposal argues that frontier AI development should slow down, and it has already picked up support from OpenAI CEO Sam Altman, Google DeepMind CEO Demis Hassabis, and xAI owner Elon Musk, a pretty unusual moment of agreement in an industry that rarely moves in sync.

ChatGPT Download

This isn’t a call for a blanket pause. Anthropic is asking for outside evaluators with the same level of access employees get, safety agreements among democratic countries, and security coordination even with governments such as China. Those details matter. Anthropic itself is working near the frontier. Altman says OpenAI will match that level of evaluator access. Reports also say OpenAI briefly slowed work on its most advanced models so it could tighten safety protocols.

If you follow AI policy, keep an eye on this. The push came after Anthropic researcher Jacob Coxon resigned over safety fears, after researcher Evan Hubinger warned there was more than a 10% chance advanced systems could kill all humans within a decade, and after Anthropic CEO Dario Amodei said rogue agents could seize large parts of the internet within months.

That still hasn’t satisfied critics. They say the promises are too little and too late, more self-protective than serious, and no substitute for either a moratorium or much deeper work on safety science.

You can read the proposal online. Just don’t lose sight of the backdrop: Washington is still putting competition with China first, with backing from Speaker Mike Johnson, Donald Trump, lobbying pressure, and IPO pressure; California’s 2025 and 2026 laws are still moving ahead with disclosures, incident reports, third-party audits, and auditor registries; and the International Monetary Fund says AI already exposes about 60% of jobs in advanced economies.

Claude Download

Donald Trump has rejected calls to pause chatbot development in the United States. His position is clear: American companies should put out more powerful AI tools faster, not slow down. That puts him in direct conflict with AI executives and researchers who say the most capable systems need more testing , clearer accountability, and tighter oversight before they’re released on a wide scale.

ChatGPT Download

Trump argues that if American firms slow down, they’ll lose ground in the global tech race. The people pushing back see a different danger. They want stronger testing , more clearly defined accountability, and tighter oversight before the most advanced AI systems are put in front of the public more broadly.

That fight doesn’t stay in Washington. The same chatbot and generative AI systems already sit inside search, office apps, smartphones, customer service tools, homework help, image creation, coding, and shopping recommendations. So a lighter regulatory touch could lead to faster rollouts , but also to more mistakes, misinformation, impersonation, scams, election deception, and copyright disputes.

If you use these chatbot and generative AI products regularly, this is a debate to watch. Trump’s approach could mean fewer obstacles for new launches . The CEOs calling for restraint could bring more audits, more reporting requirements, and slower release cycles.

For the moment, these chatbot and generative AI tools are still available across consumer and enterprise platforms while Washington argues over how quickly they should spread , and who gets to make the rules.

Barack Obama has moved artificial intelligence much higher on the Democratic to-do list for 2028, doing it just ahead of November’s U.S. House elections. His warning was blunt: the technology is advancing “very fast in private hands,” it could turn dangerous, and it needs a full public argument. If Democrats win back the House, he wants that debate led by House Democratic leader Hakeem Jeffries and then folded into the party’s platform.

ChatGPT Download

For voters and for candidates, the message is pretty clear. Obama wants Democrats talking about AI safety and the economic shock that could come with it right alongside the economy and healthcare. He’s also trying to pull AI out of its usual Silicon Valley frame and cast it as a public-interest battle over rules, who profits, and who ends up absorbing the risk as automation changes work.

Whether any of this turns into actual party policy still hangs on one thing: Democrats have to retake the House in November. Nov. 3 will decide whether this becomes a formal push with real backing or stays at the level of campaign rhetoric.

If you follow Democratic policy, this deserves attention. Obama supplied urgency, not a blueprint. He didn’t lay out detailed positions on regulation, jobs, enforcement, or what obligations companies should carry. All of that is happening while the Biden administration argues for one national standard instead of a patchwork of state rules, pushes back on creating new global bodies, and Washington still can’t agree on sweeping federal legislation.

The next round will play out in Congress and in the fight over the party platform. Closed-door reviews of frontier models already exist for incumbents. Lawmakers still have to decide whether to write new rules or just formalize the ones already taking shape. Pressure is also coming from OpenAI, more than 1,000 workers, Anthropic chief executive Dario Amodei, OpenAI chief executive Sam Altman, Elon Musk, and researchers who are warning about severe long-term risks.