15 June 2026

Banned From the Future

Keys to AI Technology and the Future of European Citizenship

With the latest US Department of Commerce Directive banning foreigners from access to Mythos and Fable 5 – the most advanced Anthropic AI models – citizenship has acquired a new function globally. On top of the traditional function of policing access to territory, confining those in possession of second-rate citizenships to the spaces of no opportunity, the legal status of citizenship can now exclude from productivity and vital technology: policing access to the future.

Make no mistake, limitations of access to knowledge and opportunities based on citizenship are as old as the concept itself, from global migration to university admissions and building capabilities for making an imprint, including running key companies. While virtually all the borders are open for Americans, Europeans and Israelis, dreaming of crossing the same for Iraqis and Kyrgyzstanis is often a crime. The arbitrary blood-based division of the world into castes marked by citizenship has long been the pillar of the totality of global distribution of rights and duties, lying at the heart of the passport apartheid. Exclusion of the majority of the global population from the spaces of opportunity has been the core function of citizenship as a colonial legal status, what Tully, Boatcă, Soomro, and others theorized, going hand in hand with racism, sexism and ethnic as well as religious discrimination: remember Einstein joining the Institute of Advanced Study, rather than Princeton University, because Princeton already had “its Jew”. And that’s a Nobel prize winner – minority students, women and foreigners faced infinitely more difficulties in the process. Banning Afghans, Burmese, Cameroonians and the Sudanese from student visas in the UK or mass slaughter of racialised foreigners in the Mediterranean by the EU are thus just the tips of the iceberg.

European continental nations, as well as the US and the UK and its dominions have always been at the forefront of the process of deploying citizenship as a sure legal tool of exclusion: from racist policies in Australia, Canada and South Africa to the mutual learning and inspiration between “the West” including US inspiration for Nazi Germany analyzed by Whitman con brio or well-known roots of European “dignity” in Nazi “honour”. The “darker legacies” are not necessarily in the past. What has remained unchanged until Trump’s last order is that “the West” has always been at the forefront of deploying citizenship at the expense of “others” – the Chinese, the Japanese, the Africans, the Communists: you name it. The Trump administration has broken this shameful trend with far-reaching consequences for the operation of citizenship as a legal status in the future.

The Border That Travels with You

Although the directive of the US Department of Commerce (which has not been made public) is described as an export regulation, its effects are anything but territorial, and therein lies its novelty. A conventional export control polices a frontier: it waits for a good, a file, or a body to cross a border before it bites. This measure does nothing of the sort. Its operative criterion is not where one stands but what one is, a citizen or an alien, so that the line of exclusion is drawn not at the edge of the polity but straight through its interior. The legal architecture that makes this possible is the “deemed export” doctrine, under which granting a foreign national access to controlled technology or source code counts, in law, as an export to that person’s country of nationality even when nothing whatsoever leaves US soil and the transaction unfolds entirely in Palo Alto or New York. The border, in other words, is internalized; it ceases to sit on the map and travels instead with the person of the foreigner, inscribed on her passport. Consider the Iranian graduate student in California, the French researcher in Anthropic’s own offices, or the Chinese or Indian engineer who helped build the very models now withheld from her.

The models at the center of the storm are not ordinary chatbots. Mythos 5, the more powerful of the two, is a frontier system of extraordinary capability, most strikingly in the domain of cybersecurity: it can read vast codebases and surface software vulnerabilities that have lain undiscovered for decades, a power that defenders of banks, hospitals and electricity grids have already begun to rely upon to harden their systems, and which, turned the other way, could serve as a formidable cyber-weapon. It is precisely this dual-use quality that led Anthropic to ration Mythos through a vetted “trusted access” program and to release to the wider public only Fable 5, a derivative model wrapped in safeguards meant to wall off its most dangerous cyber and biological capacities.

To hold such tools is to hold a general-purpose instrument of productivity, scientific discovery and economic advantage; to be cut off from them is to be relegated, quite literally, to a slower and less prosperous future, left to compete against rivals who think, code and defend themselves at a speed one cannot match. And here lies the deeper trap for the excluded: the comforting notion that Europe might simply build its own is, at least for the foreseeable future, a fiction. Frontier models of this caliber are the product of a concentration of resources that today exists essentially nowhere outside the United States (and potentially China): the colossal pools of capital (Anthropic recently raised 65 Billion USD), the scarce and largely American-controlled supply of advanced chips and the hyperscale data-centers that run on them, and a density of elite research talent that the United States has spent a decade vacuuming up from across the world, Europe very much included. The European Union has no laboratory anywhere near this frontier; its most prominent champion, Mistral, trails the leaders by a wide margin, and its compute, its cloud and increasingly its capital remain dependent on American firms. The uncomfortable implication is that Europe cannot, at present, route around the exclusion by replicating the technology; it can only hope to be let back in.

None of this is to deny that the models pose genuine dangers, or that a state may legitimately worry about a system able to manufacture zero-day cybersecurity exploits at scale; the national-security stakes of frontier AI are real. The point is that citizenship is a spectacularly bad instrument for managing these risks. The hazard inheres in the capability and in its misuse, not in the passport of the person at the keyboard: a malicious American can weaponize a vulnerability scanner as readily as any foreigner, while the French safety researcher in Anthropic’s own labs poses no threat at all, save the threat of being useful. The supposed trigger only sharpens the point. By Anthropic’s own account the “jailbreak” was narrow and equally obtainable from other deployed systems, including OpenAI’s, so the capability is already diffusing across the field irrespective of anyone’s nationality; and in a domain where there is every reason to think that performance gaps between rival models are very short-lived, the notion that one can wall off a capacity by inspecting one’s documents is a fantasy. Real security, if that is the aim, lives in instruments keyed to the risk itself: hardened safeguards, monitoring, vetted access for genuinely dangerous uses, and the transparent, fact-grounded process that the directive conspicuously lacks. To reach instead for the blunt symbolism of the passport is to mistake a caste filter for a security measure.

The Tool Turns on Its Makers

Today, for the first time, the proverbial “dreamworlds of race” propped by the legal tool of citizenship as part of the colonization and post-colonial domination effort have been shattered: excluded are not only the non-Western “others”, presumed below the “standard of civilization”. Equally, next to the Indians, Africans and Chinese, steep exclusion from technology and the possibility of building the future concerns Europeans, Australians, Canadians and all others not in possession of US citizenship, including, as noted by Time, also “Anthropic’s own employees,” the creators of the new models: top-notch computer engineering and LLM research does not know citizenship boundaries – yet, that is. For the first time since citizenship emerged as a legal way to bring down the “non-Western” populations globally, it is turning against its key creators: the Europeans. European Citizenship, unless accompanied by a US one, is becoming a ticket for sure exclusion from the technological cutting edge. And yes, once again, citizenship has always been such a tool of exclusion, only Europeans have never expected it to work against them – only against all the others. The potential consequences of the alarm bell are truly far-reaching for the nature of the citizenship of the future and the standing of European citizenship in the global hierarchy of worthy legal statuses.


SUGGESTED CITATION  Abiri, Gilad; Kochenov, Dimitry Vladimirovich: Banned From the Future: Keys to AI Technology and the Future of European Citizenship, VerfBlog, 2026/6/15, https://verfassungsblog.de/anthropic-mythos-ban-foreigners-citizenship/.

Leave A Comment

WRITE A COMMENT

1. We welcome your comments but you do so as our guest. Please note that we will exercise our property rights to make sure that Verfassungsblog remains a safe and attractive place for everyone. Your comment will not appear immediately but will be moderated by us. Just as with posts, we make a choice. That means not all submitted comments will be published.

2. We expect comments to be matter-of-fact, on-topic and free of sarcasm, innuendo and ad personam arguments.

3. Racist, sexist and otherwise discriminatory comments will not be published.

4. Comments under pseudonym are allowed but a valid email address is obligatory. The use of more than one pseudonym is not allowed.




Explore posts related to this:
AI, Anthropic, Citizenship, Donald Trump, Industrial policy, Racism