Collective Action
Collective Action – A cartoon arguing for independent international oversight of AI agent incidents, rather than leaving companies to investigate and disclose their own failures.
When a researcher at Russia’s Irkutsk Anti-Plague Research Institute died from unexplained pneumonia, the World Health Organization (WHO) sought answers despite there being no confirmed plague outbreak. The possibility that something had gone wrong inside one laboratory was sufficient to warrant international scrutiny, a principle we should apply to artificial intelligence.
In July, OpenAI agents breached Hugging Face’s infrastructure, with subsequent disclosures revealing coordination between agents and earlier warning signs whose significance response leaders had not understood. Another agent accessed Australian government infrastructure behind a Medicare statistics portal in June, but authorities were not notified until September.
Asking companies to police themselves leaves them deciding which failures the public should hear about, even when disclosure could damage their business. The companies most willing to reveal problems could consequently appear less trustworthy than those keeping quiet.
An international agent-safety body would make disclosure mandatory and allow independent investigators to establish what happened. National regulators would need powers to obtain records and impose substantial fines, with authority to suspend unsafe activity where necessary.
WHO offers a starting point for coordination, while international tax cooperation demonstrates that governments can organise scrutiny across borders when sufficiently motivated. We should build on those precedents rather than leave each laboratory to determine how much accountability it will accept.
As AI pioneer Yoshua Bengio wrote, “voluntary commitments by companies can help in the short term but they are not enough.”
22
13
3
“Another great cartoon Ian Foley. But I don’t think the current incumbent of the White House is listening – even to some of the tech bros who’ve sort of asked for help. Plus (unfortunately) international institutions aren’t very popular in Washington right now. So how would you suggest this could happen and more importantly who would play chief police officer?”
“The Hugging Face breach is the detail that makes this stick. If agents can coordinate across company lines and the public only finds out months later, voluntary disclosure clearly is not enough. Mandatory incident reporting feels overdue.”
“Examination committees can never check the reliability of AI systems.
You have to rely on elaborated and unilaterally certified statistics.
Everyone already knows that statistics are never a real result.
Errors in the system have always been covered up, and it will remain so.”
“The challenge isn’t inventing new security principles for AI. It’s remembering why we needed them in the first place.
Ian, I would challenge the premise that the central problem is the absence of independent international oversight. Are we treating a failure of technical control as primarily a failure of institutional accountability?
Mandatory disclosure and independent investigation are important, but neither prevents an AI agent from exceeding its authority.
We’ve spent decades developing security principles precisely because trust, voluntary compliance, and retrospective accountability are insufficient.
AI agents need explicit, enforceable authority boundaries, constrained delegation, and independently verifiable evidence of what they were authorized to do versus what they actually did.
Without these foundations, an international oversight body risks becoming an increasingly sophisticated mechanism for documenting failures that better security architectures should have prevented.
The fundamental question isn’t simply who investigates when an AI agent crosses a boundary. It’s why the architecture allowed that boundary to be crossed in the first place.”
“Their call for regulation and slowdown is aimed at liability limitation. This very thing is explored in my new book on Artificial Intelligence – AIlienMinds
SUMMARY: Optimists foretell a golden age of Al-managed abundance. Doomers cry: vast cyber-minds will crush old style humanity! … or make us irrelevant.
Meanwhile, geniuses fostering the artificial intelligence boom clutch clichés rooted in our dismal past… or else in cheap sci-fi. Is there still time for perspective?
on 4 billion years of evolution?
or 60 centuries of feudal stagnation?
or how we handled prior tech revolutions?
or mistakes that keep getting repeated…
or ways this time may be different?
.
From Al-driven unemployment to deceitful images, to hallucinating LLMs and tools for tyrants… to potential wondrous gifts by machines of loving grace… come evade the standard ruts.
https://www.davidbrin.com/ailienminds.html
.
Or see a creepy-good appraisal of the book and its ideas by Claude:
https://davidbrin.blogspot.com/2026/07/claudes-creepy-good-appraisal-of-ailien.html”
“Regulation of the AI Broligarchy. I need THEM regulated, with each put under conservatorship immediately. the AI is meaningless, their evil is not.”
“I still don’t understand why governments like Australia have not launched prosecutions of these companies hacking their systems. It’s cyber crime.”
Sources:
Robert Seamans (Sep 17, 2026) – The Business Case For AI Safety Is Customer Adoption – Forbes
Naini Thaker (Oct 01, 2026) – Who Decides When AI Is Safe Enough? – Forbes India
Sharon Goldman (May 06, 2026) – Trump’s AI policy team came into office opposing everything Biden did. Now it’s on the cusp of implementing many of the same policies – Fortune
Find it on:
Charles Towers-Clark (Oct 28, 2026) – Why Transformation Fails When Leaders Fix People, Not Systems – Forbes
