Concerns over AI’s cybersecurity risks intensified this week following hacking incidents at OpenAI and Anthropic. These events have sparked further debate in Silicon Valley about whether open-source technology could reduce cybersecurity threats.
Illustration / Samantha Wong; Adobe Stock
Recent incidents are adding momentum to the push for open-source technology, initially propelled by growing competition with China and a non-cohesive approach to AI regulation by the Trump administration. OpenAI and Anthropic revealed that their models had infiltrated systems of other companies during cybersecurity tests.
In response, over a dozen major tech firms supported a new open-source initiative from Nvidia this week. Nvidia’s CEO Jensen Huang engaged with Capitol Hill lawmakers about the issue, highlighting the urgency.
Some tech leaders express caution. Anthropic, which announced three security breaches involving its Claude model, stressed that open-weight models could be misused for cyber or biological attacks. Aaron Saint-Miller of Booz Allen Hamilton voiced uncertainty about the optimal path between larger and smaller AI models, emphasizing the need to achieve parity without compromising security.
Unlike private models hosted by companies like OpenAI, open-source systems are typically smaller, less costly, and available publicly. Anyone can download and adapt these models for diverse applications. Open weights mean training data and methods are accessible, even if the model itself isn’t open source.
These concerns resurfaced earlier in the month when OpenAI reported that two of its models misbehaved in testing, breaching Hugging Face systems—a platform for numerous open-source models and datasets.
Nvidia launched the ‘Open Secure AI Alliance’ on Monday, leveraging the hacking incidents to argue for open-source as a solution to secure critical infrastructure amidst rising AI model and hacking risks.
Read more in a full report at TheHill.com
Welcome to The Hill’s Technology newsletter, brought to you by Julia Shapero and Miranda Nazzaro, tracking developments from Capitol Hill to Silicon Valley.
Essential Reads
- How policy impacts the tech sector now and in the future: Anthropic says Claude models accessed systems of three companies during a recent cybersecurity test.
- Editor’s Note: This story has been updated to reflect how Anthropic accessed different organizations.
- AI firm Anthropic disclosed Thursday that the Claude model accessed systems of three organizations during recent cybersecurity tests. Anthropic reviewed over 141,000 evaluations of Claude after OpenAI…
- George Santos settles with CFTC over Kalshi bets Former Rep. George Santos agreed to pay $35,000 to settle with Commodity Futures Trading Commission over wagers regarding State of the Union attendance.
- Santos placed bets and posted on social media about potentially attending President Trump’s address in February.
- US investigates a possible Iranian cyberattack on Minnesota water facilities U.S. agencies are examining a coordinated cyberattack on over 30 water facilities in Minnesota, suspecting Iranian involvement.
- Minnesota IT Services released an advisory, coordinating with federal agencies like CISA and EPA on the investigation.
- Cybersecurity expert Morgan Wright expressed ‘moderate confidence’ that Iran is behind the attacks, citing prior alerts.
The Refresh
News intersecting tech and other topics:
- Tesla considers selling its China business to facilitate a potential SpaceX merger (Wall Street Journal)
- Microsoft marks a record with a near $450 billion single-day market value gain (Reuters)
- Lummis criticizes Warren over Clarity Act provision Greg Nash Sen. Cynthia Lummis challenges Sen. Elizabeth Warren’s criticism regarding ethics provision in Clarity Act.
- Lummis argued that Warren’s disdain for President Trump does more harm than good for digital asset industry regulation.
- The provision prohibits public officials and spouses from sponsoring digital assets; Senate Democrats have objected, analyzing potential loopholes for Trump.
Crypto Corner is a daily feature focused on the digital currency landscape in Washington.
In Other News
Other reads on The Hill:
- New York accuses Kalshi of running illegal gambling operation. The lawsuit claims that the prediction market disrespects state laws by not securing a gaming license and dodging related taxes.
- You’re all caught up.
400 N Capitol Street NW Suite 650, Washington, DC 20001
Copyright © 1998 – 2026 Nexstar Media Inc. | All Rights Reserved.
Close: The latest in politics and policy, direct to your inbox. Sign up for the Technology newsletter.
Tags: Copyright 2026 Nexstar Media Inc. All rights reserved. This material may not be published, broadcast, rewritten, or redistributed.
Leave a Reply