Scoop: Wes Moore to meet with AI execs on Mythos-era threat

• AI is likely to be a top issue in the 2028 presidential race, and Moore — who's widely seen as a potential contender — could benefit from positioning himself as a leader on the topic.
• Many Democrats also are eager to win over Silicon Valley CEOs after some in the tech industry broke with the party in 2024 and backed Donald Trump.

Zoom in: Anthropic revealed this month that its new model, Claude Mythos Preview, poses such massive cybersecurity risks that it can't be released to the public.

• The company said it disclosed the new model's frightening abilities to sound the alarm about a new age of heightened AI threats.
• Governments and corporations have expressed concerns about the risks raised by Mythos.

Driving the news: Moore has been holding private conversations about AI with OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei over the last year, a person close to the governor told us.

• The discussions have centered on how Maryland can benefit from AI while protecting residents from its potential harms, the source said.
• But since Mythos' recent rollout to a small number of companies to shore up security vulnerabilities, Moore has been concerned about how AI could unleash major cyberattacks — and decided to convene AI leaders in person, the source said. AI safety experts also are expected to join the conversation.
• Former Commerce Secretary Gina Raimondo also is slated to be at Monday's dinner. She and Moore have been talking recently about Mythos.

In Moore's view, the U.S. government has failed to manage AI, forcing states to step in.

• Moore also plans to speak with the AI leaders Monday about mitigating AI-driven job losses and positioning Maryland to capitalize on the economic shift brought on by the technology.

Zoom out: Anthropic says Mythos is shockingly good at finding and exploiting security flaws in software programs.

• One source briefed on coming AI models previously told Axios that a large-scale cyberattack could happen as soon as this year.

• Some important context: Moore's state is home to Fort Meade, the headquarters of the National Security Agency and Cyber Command.

What they're saying: Moore has sought to position himself as middle-of-the-road on AI.

• "He's a realist," said Ammar Moussa, a Moore spokesperson. "AI is coming, and we need to harness it. But it's going to be massively disruptive for our economy and security, and we need to take this seriously because the federal government is not."
• An Anthropic spokesperson pointed to comments it made last year about Amodei meeting with Moore and the company working with his administration on government services. Microsoft declined to comment, and OpenAI did not respond.

Catch up quick: Several other Democrats eyeing presidential bids are trying to show voters they're well-equipped to handle AI's challenges. New York Rep. Alexandria Ocasio-Cortez, for example, has proposed a moratorium on data centers until legislation protecting workers, consumers and the environment is passed.

• Meanwhile, Pennsylvania Gov. Josh Shapiro, Illinois Gov. JB Pritzker and Moore have wooed data centers to their states to try to create jobs and grow their tax bases — but they've recently taken a more measured approach as voters have soured on the projects.
• Moore announced an agreement with state lawmakers last week to support a bill that would make data centers pay for their own infrastructure upgrades.

Go deeper... Scoop: NSA using Anthropic's Mythos despite blacklist