New York Assemblymember Alex Bores asked Congress to ban recursive self-improvement, or RSI, in a post on X on November 6: “Congress should ban RSI, this year if possible”. Bores defined RSI as AI companies trying “to fully automate AI research; to allow the AIs to build the next AI”, and wrote that “OpenAI says they are on track to do so by March 2028”. X showed the post at about 499,000 views on November 8.
OpenAI’s own March 2028 goal is a “true automated AI researcher”, as Sam Altman described it. By our arithmetic, March 2028 is 16 months from November 8, 2026, and the next US general election, on November 7, 2028, is 24 months away. Bores wrote: “We don’t get another election before AI companies reach this threshold”.
Business Pill · Guardrails Explained
A short Business Engineer explainer on guardrails in AI systems. It teaches the general idea only and does not discuss Bores or recursive self-improvement.
The key insight: New York Assemblymember Alex Bores asked Congress on 6 November to ban recursive self-improvement, this year if possible, citing OpenAI’s goal of a true automated AI researcher by March 2028 and a METR report on Anthropic. OpenAI says it reached its automated research intern goal in September 2026, and Anthropic’s chief scientist told the Guardian a decision on letting AI self-improve could come between 2027 and 2030. By our arithmetic, OpenAI’s target is 16 months away and the next US general election is 24 months away.
Bores’s X profile lists the roles “NY Assemblymember, computer scientist, AI Safety Advocate”. The post also says: “Anthropic employees say they are on track to do so by 2027”. Bores quoted a METR post about a new report and wrote that “The METR report below suggests they are likely right”.
The case Bores makes is about compounding errors: “This is a bad idea given current technology. Any mistake or misalignment in current AIs could be multiplied in each generation”. Bores called the idea “a remarkably consensus opinion” and listed supporters including Bernie Sanders, Steve Bannon, Ralph Nader, Glenn Beck, Yoshua Bengio and Daniel Kokotajlo.
The post names no bill. Bores wrote: “The exact mechanism can be debated. The ban doesn’t have to be permanent either”. Rohan Paul, an AI commentator on X, shared it on November 7 with the line “RSI ban talk is getting louder”.
Stated targets for automated AI research against the US political calendar. Data: Sam Altman on X (29 Oct 2025); OpenAI (6 Sep 2026); Jared Kaplan to the Guardian (2 Dec 2025); METR; Alex Bores on X (6 Nov 2026). Our arithmetic: months from 8 Nov 2026 to March 2028 (16) and to 7 Nov 2028 (24).
What Was Posted on X
Every AI company is trying to fully automate AI research; to allow the AIs to build the next AI.
This is called Recursive Self Improvement (RSI).
OpenAI says they are on track to do so by March 2028.
Anthropic employees say they are on track to do so by 2027.
The METR report below suggests they are likely right.
This is a bad idea given current technology. Any mistake or misalignment in current AIs could be multiplied in each generation.
We don’t get another election before AI companies reach this threshold.
Congress should ban RSI, this year if possible.
This is a remarkably consensus opinion, supported in various versions by: -Bernie Sanders -Steve Bannon -Ralph Nader -Glenn Beck -Republican and Democratic attorneys general -Whistleblower Daniel Kokotajlo -Turing Award winner Yoshua Bengio -Humans First -Open Markets Institute -Ezra Klein -Arvind Narayanan (a Princeton computer scientist known for arguing that “AI is a normal technology”)
The exact mechanism can be debated. The ban doesn’t have to be permanent either (see below for both). This is not the only thing we have to do in regards to AI, far from it! But it is urgent.
New York Assemblymember Alex Bores’s post of 6 November asking Congress to ban recursive self-improvement, ‘this year if possible’. The post quotes METR’s announcement of its report on Anthropic and names no bill; this piece checks its OpenAI and Anthropic timing claims against the labs’ own statements.
Rohan Paul, an AI commentator, shared the Bores post on 7 November with the line ‘RSI ban talk is getting louder.’ The comment is Paul’s own reading; this piece reports what Bores wrote and what the labs have said.
Imagine you are put in charge of monitoring a massive water tank the size of a football stadium. The water inside doubles in volume every sixty seconds. At 12:00 noon, the tank overflows. When should you have become concerned? At 11:59 the tank was only half full.
Andrew Curran posted at 16:28 UTC on 8 November a parable about exponential growth: a stadium-sized tank whose water doubles every 60 seconds overflows at noon, yet was only half full at 11:59. The post, quoting Curran’s earlier one on why people misjudge exponential growth, is commentary and makes no claim about any lab’s timeline.
I’ve summarized where things stand at the AI frontier:
– Unreleased models (Anthropic’s Models 3 and 4) are now half-way between AI2027’s Agents-1 and 2 (graphs fixed for the bug flagged by the AI2027 authors). @DKokotajlo please let me know if I missed something.
– We’re right on track with AI2027’s prediction of an automated coder (AC) for April 2027, wild. Public models are now massively lagging unreleased models (1 day vs. 1 week of autonomous high-quality work). Fuelled by RSI, the gap is huge and growing.
– This trajectory, if continued, means a world-changing intelligence explosion in 2027, so the public needs to know. I highly recommend reading more on the topic to grasp what’s coming.
Karim Beguir posted on 8 November a summary comparing AI progress with the AI 2027 scenario. In the poster’s analysis, unreleased Anthropic models sit half-way between the scenario’s Agents-1 and Agents-2, and progress is on track for the scenario’s automated coder by April 2027. These are the poster’s own readings against a forecast scenario, not established facts.
What the Labs Have Said
Altman wrote on X on October 29, 2025: “We have set internal goals of having an automated AI research intern by September of 2026 running on hundreds of thousands of GPUs, and a true automated AI researcher by March of 2028. We may totally fail at this goal”. In a September 6, 2026 post, OpenAI said it had “now reached the goal” of an automated research intern and was “making strong progress toward creating an automated AI researcher by March of 2028”.
The same OpenAI post says that as of mid-August its research organization “uses 3.1 agent-workdays of effort for every workday of human labor”. It adds that these are reasons to build automated research capabilities, “but they do not mean that rapid RSI is necessarily an outcome we should pursue”.
Anthropic’s Anthropic Institute essay on the topic says: “We are not there yet, and recursive self-improvement is not inevitable. But it could come sooner than most institutions are prepared for”. It reports that “Anthropic engineers on average ship 8x as much code per quarter as they did from 2021-2025”. Anthropic chief scientist Jared Kaplan told the Guardian in December 2025 that the decision to let AI recursively self-improve could come between 2027 and 2030, calling it “in some ways the ultimate risk”.
What METR Found
The METR report Bores cited was published on November 5 after Anthropic gave METR access to internal data. As we reported in METR: Anthropic Likely Improving Models at 2x Its Old Rate, METR says Anthropic is “likely improving the capabilities of their models on well-specified tasks at around twice the rate they did between March 2024 and October 2025”, a finding it gives “with high uncertainty”. Our follow-up, Anthropic’s R&D rose 5.3x as METR’s progress rate rose 1.9x, compares those figures.
From a City Ban to Congress
Two days before Bores’s post, the Bloomington, Indiana, Common Council voted on November 4 to ban the distribution and deployment of superintelligent AI inside the city, as we reported in Bloomington Votes to Ban Superintelligence, Sets 95% Bar. In Congress, Rep. Yassamin Ansari called on November 4 for “a ban on artificial superintelligence”, and is one of 10 House members named on H.R. 10538, the Ban Artificial Superintelligence Act of 2026, as we reported in Ansari wants a day-1 AI pause.
Bores represents New York State Assembly District 73, a term that ends on January 1, 2027, according to Ballotpedia. Ballotpedia says Bores did not run for re-election to the Assembly in 2026 and lost the Democratic primary for New York’s 12th Congressional District on June 23, 2026.
The Business Lever: Automated Research
Both labs describe automated research as part of how they build models. OpenAI reports 3.1 agent-workdays per human workday in its research organization, and Anthropic reports 8x as much code per engineer per quarter. Bores’s proposal targets that step, the automation of AI research itself, rather than any product a customer buys.
The Structural Read
As we read the post, Bores is aiming at a step inside the labs, the automation of AI research, rather than at a product or a model release.
As we read the lab statements, both OpenAI and Anthropic already report large internal gains from automated research, and both say that further steps are not a foregone outcome.
As we read the calendar, a city ban on 4 November and a state lawmaker’s call on 6 November put RSI on the political agenda within the same week.
Alex Bores, New York Assemblymember, on X
“Congress should ban RSI, this year if possible”
Three Implications
FOR OPENAI Its stated goal is a true automated AI researcher by March 2028, and its September 2026 post reports 3.1 agent-workdays per human workday in research.
FOR ANTHROPIC Its Institute essay says recursive self-improvement is not inevitable but could come sooner than most institutions are prepared for.
FOR LAWMAKERS Bores’s post names no bill, and Ballotpedia says the Assembly term ends on 1 January 2027.
The essay says “The governance layer sits at the intersection of capability, regulation, national security, access control, and geopolitical alignment.”
As we read it, Bores is proposing to draw one line in that layer at the automation of AI research itself. The essay does not discuss Bores or a ban on recursive self-improvement.
What Is Not Established
Bores’s post names no bill, sponsor or definition of what a ban would cover. It does not name the Anthropic employees it cites for 2027; the Anthropic statement on timing we read is Kaplan’s 2027-to-2030 window for a decision. OpenAI says it “may totally fail” at its March 2028 goal, and METR gives its estimate with high uncertainty.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
New York Assemblymember Alex Bores asked Congress on November 6 to ban recursive self-improvement, “this year if possible”. OpenAI’s stated goal is a “true automated AI researcher” by March 2028, and Anthropic’s chief scientist has put a decision on letting AI self-improve between 2027 and 2030. By our arithmetic, OpenAI’s target is 16 months away and the next US general election is 24 months away.
A note on sourcing. Bores’s quotations are from the post on X of 6 November 2026, which we read in full. OpenAI’s goals and figures are from Sam Altman’s post of 29 October 2025 and OpenAI’s post of 6 September 2026; Anthropic’s from its Institute essay and Jared Kaplan’s interview with the Guardian. Bores’s Assembly term and primary result are from Ballotpedia. The month counts are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
