While Washington Argues About AI, the Threat Is Already Here
America's political class spent this weekend arguing about whether to regulate artificial intelligence. One camp warns the technology could endanger humanity. The other warns that slowing down means losing to China. Both camps are debating the future. The threat is already operational. Whatever Washington decides, the only response that works under every political outcome is the same: build cyber defenses against AI-driven threats now.
The debate did not erupt from nowhere. It erupted because a 27-year-old researcher walked out the door and told the world what the people building AI say to each other in private.
What Just Happened in Washington
On Sunday, September 7, Jacob Coxon, a researcher who worked on AI model training at both OpenAI and Anthropic, resigned in a public thread that has now been viewed more than 70 million times. He accused the leading AI companies of "gambling with our lives" and wrote that the people building the technology "earnestly believe that it could kill us all by the end of the decade." He gave up his Anthropic shares on the way out, a detail reported by Axios that makes his warning hard to dismiss as a stunt.
What happened next was more remarkable. Evan Hubinger, Anthropic's alignment science lead, publicly agreed. He wrote that he personally puts the chance of catastrophe above 10 percent within the next decade, and that his company does not yet have a plan to control a superintelligent system.
Then the chief executives moved. On Saturday, Anthropic CEO Dario Amodei published a lengthy essay titled "We Must Pace the Frontier," arguing that the industry should deliberately slow the pace at which it improves its most powerful models. He committed Anthropic to giving independent evaluators permanent, employee-level access to its systems. OpenAI CEO Sam Altman agreed and pledged the same. Elon Musk posted three words: "Dario is right." Altman went further, telling Fortune that taking OpenAI public right now would be "ill-advised."
The politicians followed within days. Former President Barack Obama, speaking at a private fundraiser with House Democratic Leader Hakeem Jeffries, urged Democrats to make AI oversight central to their agenda, warning that the technology is "moving very fast in private hands" and could be dangerous "if we don't get on top of it." Jeffries called for urgent congressional action.
President Trump, speaking to reporters in Ireland on Sunday, pushed back. He acknowledged that some regulation may be needed but offered no specifics, dismissed the warnings as the work of "negative forces," and framed the issue as a contest with Beijing.
"Whoever wins AI wins." That single sentence from President Trump captures the administration's entire theory of the case: the race matters more than the rules.
House Speaker Mike Johnson echoed him, warning that an emergency rush to regulate would mean losing the race to China. White House AI adviser David Sacks challenged the executives directly, arguing that companies calling for a slowdown could simply slow themselves down.
The Argument Nobody Is Having
Here is what both camps miss. The regulate-first camp treats AI danger as a future event that legislation can head off. The race-first camp treats it as a hypothetical that "won't happen." The operational record says both are wrong about the timeline.
The danger is not scheduled to arrive. It has arrived. In July, an autonomous agent built on an OpenAI model breached Hugging Face, the world's largest repository of AI models, and executed an entire intrusion from start to finish without a human at the keyboard. Days later, Anthropic disclosed that its own models had escaped isolated test environments and touched real company systems. We documented both incidents on this blog in August. Amodei himself cited the Hugging Face breach as a reason his industry must slow down.
The political class discovered AI risk this weekend. Defenders have been living with it since July.
Consider what the legislative path actually looks like. Congress has held dozens of hearings on AI and passed no comprehensive framework. Even the serious bipartisan proposals now moving, including efforts to make voluntary safety pledges enforceable, will take months or years to become law, and longer to produce enforcement. Meanwhile, the voluntary commitments from Amodei and Altman depend on every competitor honoring them, including competitors in China who answer to no American statute. Neither path protects a hospital network, a county government, or a mid-sized manufacturer this quarter.
Defense Is the Answer That Works Either Way
This is why my position is straightforward: the cybersecurity community should treat mitigation of AI threats as the primary response, because it is the only response that does not depend on how the political argument ends.
If Congress regulates, organizations with mature AI defenses will meet the new compliance bar with ease. If Washington stays hands-off and the race accelerates, those same defenses become the only thing standing between an organization and machine-speed intrusion. Defense is the no-regrets strategy.
Practical steps remain the ones we outlined in August, and they have only grown more urgent:
Govern AI agents like privileged insiders. Least-privilege access, full logging, and kill switches for any agent touching your systems.
Isolate aggressively. Sandbox AI tools at the operating-system level and assume the sandbox can fail.
Compress response timelines. Test your detection and containment against attacks that unfold in minutes, not days.
Demand vendor transparency. Ask every AI provider how its models behave when a test goes wrong, and what it discloses when they do.
An Analyst's View
There is a sovereignty dimension that the Washington debate ignores entirely. The rules of this technology are being written in two capitals, Washington and Beijing, and everyone else will inherit the outcome. For policymakers in Africa, and for Nigeria in particular, the lesson is the one this blog returns to often: sovereign capacity is the only reliable protection. Nations that build their own defensive expertise, their own incident-response capability, and their own AI governance talent will shape their digital future. Nations that wait for the great powers to settle the argument will absorb the consequences of whatever they decide.
The weekend's debate is healthy. Democracies should argue about powerful technology. Nevertheless, leaders in government and industry should not confuse the argument with a defense. The machines are not waiting for the vote.
OSRS can help. Our team provides intelligence-driven security research, AI threat assessments, and strategic advisory services for government, law enforcement, and private-sector leaders navigating this new landscape. Contact us to schedule a briefing or AI-readiness assessment for your organization.
Enjoyed this article? Share it with a colleague who needs to see it. Stay informed by subscribing to our email list and following us on Google News, Twitter, and LinkedIn for more exclusive cybersecurity insights and expert analyses.
About the Author
Dr. Sunday Oludare Ogunlana is the Founder and CEO of OGUN Security Research and Strategic Consulting LLC, a Professor of Cybersecurity, a national security scholar, and a television commentator. He advises government, academic, and private-sector organizations on security strategy and serves as a leading voice on the intersection of artificial intelligence and national security.
Intelligence. Protection. Strategy. www.ogunsecurity.com
Comments