The United Kingdom’s AI Security Institute has revealed that advanced artificial intelligence models can compromise corporate networks with alarming frequency, exposing a critical vulnerability in national infrastructure. Anthropic’s Mythos model successfully took over corporate networks in six out of 10 attempts during testing, while OpenAI’s GPT-5.5 achieved the same in three out of 10 tries. This rapid advancement in cyber capabilities, confirmed by British AI Minister Kanishka Narayan, signals an accelerating threat to the digital security landscape, with profound implications for national sovereignty and the safety of citizens.
Nine of the nation’s top cyber researchers and tech leaders, who have experimented with Mythos and GPT-5.5 in controlled settings, concluded that these tools are advancing much faster than anticipated. Lee Klarich, chief product and technology officer at cybersecurity company Palo Alto Networks, described testing Mythos as a "game-changer," stating it was "more powerful than I thought it was going to be." Isaac Evans, CEO of cybersecurity company Semgrep, noted Mythos "exceeded our expectations," demonstrating an "uncanny ability around exploit generation" and being capable of generating "a SolarWinds every quarter." The 2020 SolarWinds breach, attributed to the Russian government, affected over 18,000 organizations and is considered one of history's worst hacks.
Unleashed Power, Unchecked Threat
Jonathan Trull, chief information security officer of IT security company Qualys, which is testing GPT-5.5, stated the model "can basically do what your most advanced app security engineer can do." Cloudflare Chief Security Officer Grant Bourzikas confirmed Mythos can both identify vulnerabilities and write code to exploit them, marking a "real step forward" for advanced AI. Broadcom, after testing Mythos against its own software code, described its findings as "jolting" in a report published last month, adding, "We are learning things that appear unlikely to ever have been uncovered by human researchers alone."
Anthropic itself warned last month that Mythos had "already found thousands of high-severity vulnerabilities, including some in every major operating system and web browser," and cautioned that the consequences of deploying such technology could be "severe" for global economies, public safety, and national security. The companies developing these frontier AI models have kept testing limited to small groups of "trusted organizations," maintaining control over tools that have outpaced publicly available digital resources and even the most skilled human minds. This concentration of power in private hands raises questions about national control over critical defense capabilities.
National Security Under Siege
Government agencies, congressional committees, banks, and regulators have been "clamoring for access" to these advanced AI tools in recent weeks, seeking to secure critical networks before adversaries acquire the technology for devastating cyberattacks. Concerns are escalating that China and other foreign adversaries could soon develop their own advanced AI tools, with China already engaged in an "industrial-scale campaign to copy American AI technology" through "distillation attacks." This aggressive pursuit by rival powers underscores the immediate threat to national security and the imperative for sovereign defense.
The Trump administration is described as "acutely aware of these dangers" and is "scrambling" to collaborate with tech companies, government agencies, and critical infrastructure groups to deploy these tools "quickly and safely." However, President Donald Trump abruptly postponed signing an executive order "earlier this week" that would have established a voluntary process for tech companies to submit certain AI models to the federal government for testing. This delay injects uncertainty into national efforts to secure vital systems.
Elite Interests Over National Defense
Former AI czar David Sacks raised concerns about the executive order "stifling innovation" with President Trump at the last minute, plunging the regulatory process into "chaos." President Trump told POLITICO "Friday" that he had "many" concerns about the draft executive order and worried it was "inhibiting the industry." This prioritization of corporate "innovation" over immediate national security concerns highlights a dangerous alignment of elite interests that could leave the nation vulnerable.
While advanced AI tools could potentially aid cyber officials in securing critical networks across water facilities, hospitals, and telecommunication systems, with one expert suggesting a "future state where we will actually be producing more secure products," Isaac Evans of Semgrep offered a stark counterpoint. Evans stated, "These model developments mainly are advantages for attackers rather than defenders," reinforcing the immediate and growing threat to national infrastructure and the people it serves. The ultimate cost of this unchecked technological advancement, driven by private enterprise and influenced by elite lobbying, remains to be borne by the native population.