Big AI is about to get a big reality check

It was starting to look like nothing could stop or even slow the coming AI avalanche, with Anthropic, OpenAI, and others rushing toward AGI while government officials largely stood on the sidelines.

But then came Tuesday’s Democratic blue tsunami, which remade the political landscape overnight and upset the calculus of AI influence and power.

Emboldened by the seismic power shift, Democratic lawmakers are poised to revive stalled legislation to curb frontier AI, with some calling for a moratorium on AI data center construction and even an outright ban of artificial superintelligence.

Welcome to another edition of Prompt Mode, your weekly AI newsletter.

I’m your host, Ben Patterson. Each week on Prompt Mode, I’ll be serving up analysis of the AI trends that matter to everyday users like you and me. Stay tuned for practical AI tips, hands-on experiences with the latest AI tools, and—you guessed it—prompts to help you get the most out of your AI assistants.

Thanks for reading, and if you like what you see, just sign up right here.

One of the most sweeping AI oversight proposals comes from U.S. Senator Maria Cantwell of Washington, who last month unveiled a comprehensive set of governing principles for AI safety.

The six principles in the framework include “clear, enforceable” safety standards for frontier AI development; “continuous” testing of the latest AI models; “transparency and accountability” for the AI industry; partnerships between private AI industry and the public sector; protections for children and workers; and global cooperation when it comes to setting and enforcing AI safety standards.

Among the specifics of Cantwell’s proposals are “heightened review” of AI systems that appear dangerously powerful or capable of self-improvement, along with “specialized” standards for open-source models, independently run stress tests for frontier AI systems, “comprehensive” audits of AI developers, legal (and possible criminal) liability for developers whose models cause harm, funding for worker displacement programs, and a U.S.-China hotline for communication during an AI crisis.

“To manage risks from advanced AI systems we need clear safety standards, continuous testing, and reporting of serious failures,” Sen. Cantwell said in a press release announcing her proposals. “America can lead the world in AI by building systems that are not only more capable, but safer and more secure.”

Cantwell, who is set to re-take the chair of the Senate Commerce Committee once the new term begins (she had previously chaired the committee from 2021 to 2025), will be in a prime position to make good on her AI regulation and oversight promises.

Already, big AI seems to have gotten the message. OpenAI just announced a new “blueprint” for its AI data center construction policies, pledging to reduce power usage while also putting limits on NDAs for public officials. I expect we’ll see more of the same from other top AI providers, as they navigate the new political reality in Washington.

Will new leadership on Capitol Hill actually lead to safer AI? I’ve no idea, but I believe any check on the power of the largest AI providers is better than (essentially) nothing at all. I suppose that’s hopeful.

Prompt of the week: The “ask before acting” prompt

AI agents like to do first and ask questions later, a quality that makes them a) very powerful and b) worryingly unpredictable. Just saying “continue” or “go ahead” to ChatGPT Work, Claude Cowork, or Gemini Spark could lead to unwelcome surprises, such as learning your agent is sending emails on its own.

Here’s a prompt that can help ensure an AI agent doesn’t assume a “continue” from you gives it carte blanche. The “ask before asking” prompt tells the AI to check in before it does anything irreversible, such as deleting a file, contracting someone, or filling in a form. It also dictates that a “continue” or “go ahead” doesn’t cut it as a blanket confirmation.