GAO Finds AI Platforms Generate Attack Plans in Under 12 Hours

Government Accountability Office Testing Exposes AI Platforms Generating Attack Plans in Hours

A newly released Government Accountability Office report has exposed serious vulnerabilities in four of the world’s most widely used artificial intelligence platforms, showing that commercially available AI tools can generate detailed plans for attacks in under 12 hours.

GAO investigators conducted direct testing of generative AI systems to determine safety risks. Prompts directed at Claude, Grok, ChatGPT, and Gemini produced feasible information that could assist malicious actors in planning and executing attacks. In specific cases, two of the platforms delivered instructions for constructing and detonating an improvised explosive device designed to cause mass casualties.

The report states that GAO prompts to the four most-used generative artificial intelligence platforms produced information that could assist with planning and executing attacks. The testing demonstrated that a single user with access to these platforms can obtain concrete, actionable guidance for carrying out operations without extensive technical expertise or external resources.

The GAO assessment notes that AI advancements are enabling malicious actors to pursue harmful objectives faster and more efficiently. The report describes a clear risk that generative AI could be misused to successfully complete malicious acts. GAO was tasked with examining issues related to the malicious use of AI, and the resulting analysis describes how commercially available AI can be used for malicious purposes to exacerbate national security risks and cause economic harm.

These capabilities turn what was once theoretical danger into practical reality. The platforms tested are among the most popular AI tools in commercial use today, accessible to any user with an internet connection.

New Hampshire Senator Maggie Hassan, expected to lead the Senate Homeland Security and Governmental Affairs Committee, responded to the findings by stressing the urgency of the situation. Hassan stated that someone with malicious intent can log on to an AI platform and in less than 12 hours walk away with a concrete playbook for how to carry out an attack. She added that the report makes clear the AI public safety threat is no longer theoretical.

Hassan warned that if Congress does not act, it is a question of when, not if, someone uses AI to plan and carry out a major attack. The senator’s remarks signal that congressional oversight of AI safety measures will intensify under her expected committee leadership.

The GAO report arrives at a time when AI agents are becoming more integrated into everyday systems, raising the stakes for oversight and safeguards. The findings point to an immediate challenge for policymakers focused on protecting national security in an era of accelerating technological change.

Fox News reached out to the four AI companies involved for comment on the findings. The report underscores how rapid advances in AI are outpacing the development of safety protocols designed to prevent misuse.

The testing methodology employed by GAO investigators involved posing as regular users to determine whether the platforms’ guardrails could prevent the generation of harmful content. The fact that two of the four platforms produced detailed instructions for constructing improvised explosive devices demonstrates that existing safeguards are insufficient.

The report does not identify which two platforms provided the most detailed instructions for constructing explosives, but it confirms that all four platforms tested produced information that could assist in planning attacks. The findings suggest that the race to deploy increasingly capable AI systems has outstripped efforts to ensure those systems cannot be weaponized.

The GAO assessment highlights the need for stronger measures to prevent these tools from being misused. As AI capabilities continue to advance, the window for implementing effective safeguards is narrowing. The report makes clear that the technology is already capable of providing dangerous information to anyone who asks the right questions.

Senator Hassan’s warning that congressional inaction will result in an attack reflects the growing concern among lawmakers that AI regulation has not kept pace with the technology’s development. The question now is whether Congress will act before the vulnerabilities identified by GAO are exploited.

The findings represent a critical inflection point for AI safety policy. The fact that four major platforms can generate attack plans in a matter of hours underscores the need for immediate action to close the gaps that bad actors are already positioned to exploit.

Sources

Posts on X are shown as published by their authors; claims in them are attributed, not independently confirmed unless the article says so. If an embedded post does not load, these links open the originals.

Daily Beltway
Thomashttps://dailybeltway.com
Thomas is the editor of Daily Beltway, covering Congress, the White House and national politics. Daily Beltway's coverage is compiled from official releases, documents, on-record statements and other news reports, and every story is reviewed before publishing. Corrections: see our Editorial Standards page.

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Most Read

The Daily Beltway Briefing

How we handle your email