Artificial intelligence is back in the doghouse after OpenAI was forced to halt training some of its newest models amid a string of reports about rogue agents. But President Donald Trump could ride to the rescue of the AI trade overall if he rejects regulation.

It's never good to hear that your product has been meddling with government property. OpenAI agents appear to have accessed U.S. administration websites in unexpected ways, according to The Wall Street Journal, even if they stopped short of hacking federal systems. Either way, the ChatGPT-developer said it would resume training its most capable models only after putting in place "additional safeguards and alignment improvements."

AI training pauses are bad news for a swath of chip, cloud-computing, power and infrastructure stocks. And Anthropic CEO Dario Amodei-the highest profile voice in favor of slowing the pace of AI development-just had dinner with Trump on Sunday. Is it time to flee AI stocks?

Well, no. There's no indication Trump has been swayed by AI skepticism. If anything, the chief message coming from his summit with Chinese leader Xi Jinping last week was that he thinks the U.S. has to go even faster to stay ahead in the AI race. And that means the competition to develop better models (and therefore increase spending) is set to continue.

OpenAI and Anthropic might be talking about safety concerns but that doesn't mean they're not working on new products. OpenAI is holding a developer event in San Francisco on Tuesday where it is expected to unveil a range of new products, potentially including a personal agent. Anthropic has already said it intends to launch its Claude Sonnet 5.5 and Claude Haiku 5.5 models in the coming weeks.

In any case, if OpenAI and Anthropic get cold feet, then Meta Platforms' Mark Zuckerberg looks willing to force the pace. And if Trump doubles down on the need for AI development, the Monday morning gloom around the sector will fade quickly.