A few weeks ago, Fortune’s Beatrice Nolan was the first to report that a powerful new AI model called Mythos was in development at Anthropic. Her reporting was a great scoop, but it also surfaced serious concerns about the large language model’s potential to enable uncontrollable cybersecurity risks.

Since Nolan’s story broke, those worries have rapidly spread. Unlike prior models it had released publicly, Anthropic opted to give Mythos to a small number of companies for quiet testing. It is also planning to tightly restrict the model’s release, precisely because of the cyber risks.

One Fortune 500 executive I spoke with recently told me their team has been testing the model. They’re concluding that both the hype and the security risks are very real.

Nolan gives a good assessment of what’s at stake: “Attackers, even those lacking high-level technical skills, can now launch highly automated attacks across thousands of systems at once in a massive, coordinated assault,” she writes. “Unless government and industry harden defenses, the world could see a wave of devastating cyberattacks taking down banking systems, power grids, hospitals, or water systems. It is exactly such a nightmare scenario that Anthropic says it is hoping to head off by limiting Mythos’s release.”

Those with access to Mythos say the model is identifying vulnerabilities faster than companies can patch them. The fix, Fortune’s Sharon Goldman writes, requires a mindset shift: Protections need to be built into software from the ground up, rather than handled after the fact.

Mythos is proving consequential on other fronts, too. Fear, it turns out, is a powerful marketing tool, and Anthropic appears to be positioning itself as the safe, responsible solution to the very risks it’s surfacing. (“Safe AI” has always been a big part of Anthropic’s brand.) The model may also be key to repairing Anthropic’s relationship with the White House after some very expensive tension with the Defense Department. Axios reports that Anthropic CEO Dario Amodei traveled to Washington to meet with White House Chief of Staff Susie Wiles, with Mythos and its security implications as the basis for the conversation.

For more on Mythos and other evolving AI stories, read recent reporting from Beatrice Nolan and Sharon Goldman.

In other major business news this week, Fortune’s Shawn Tully got sources to open up about a potential merger of United Airlines (No. 74) and American Airlines (No. 81) that could reshape the U.S.’s global positioning in the skies. His takeaway: The monster deal is by no means impossible.

Follow Alyson on X, LinkedIn, TikTok, Instagram, and the Titans and Disruptors vodcast.