Federal agencies will shift to a priority and risk-based method of logging cybersecurity events under a Friday memo from the Office of Management and Budget aimed at cutting “red tape” and costs. The memo from OMB Director Russell Vought rescinds and replaces a previous directive from the Biden administration issued after the 2020 SolarWinds breach that affected both the public and private sectors. While the previous policy “improved foundational capabilities across agencies,” OMB said the amount of data agencies were required to retain was costly and operationally difficult. In its place, the Trump directive outlines “a risk-based, prioritized logging approach” to logging. OMB’s policy comes amid concern about the use of artificial intelligence and automation to fuel cyberattacks. That technology can speed up the process of gaining access to a system and help covertly maintain that access for a long time. It’s also increasingly being used by threat actors, the memo said.

Anthropic’s Mythos large language model is the talk of federal tech and cyber practitioners across the Beltway, and for good reason. According to the company, its month-old Project Glasswing initiative, which allows select researchers to get their hands on the Mythos model, has uncovered more than 10,000 high- or critical-severity software vulnerabilities across systemically important code, a finding that Anthropic says has shifted the central problem in cybersecurity from discovering flaws to verifying and patching them. The findings, drawn from partner reports and independent evaluations, mark one of the first large-scale accountings of what a frontier AI model can do when pointed at widely used code, and of the bottlenecks that emerge once it does. Several partners reported that their rates of bug discovery had increased more than tenfold.

The Daily Scoop Podcast is available every Monday-Friday afternoon.

If you want to hear more of the latest from Washington, subscribe to The Daily Scoop Podcast  on Apple Podcasts, Soundcloud, Spotify and YouTube.