SIAgentNews.com
← All news

Anthropic IPO faces protests over governance and AI safety risks

Protesters and advocacy groups are challenging Anthropic's public debut, citing concerns over founder control, healthcare bias, and recursive self-improvement.

· First reported by Unite.AI

Cinematic film still: picture for the story 'Anthropic IPO faces protests over governance and AI safety risks'
AI-made cinematic film still for this story. No real people or products are shown.
50.1%Founder voting power

The short version

  • Anthropic's IPO prospectus reveals a single Class F share held by founders carries 50.1% of the total voting power.
  • Protesters from ACT UP NY and Ban RSI are demanding verifiable steps against recursive self-improvement in AI development.
  • Advocates are calling for a boycott of the IPO to prevent the use of retirement savings and taxpayer money for AI investment.
  • The company frames AI-assisted development as a structural advantage rather than a safety risk in its official filings.

Why it matters

This story highlights the tension between the rapid commercialization of AI and the public demand for safety guardrails. It also raises questions about how much power individual founders will retain over technologies that have significant societal impacts.

Public protests at the IPO

Anthropic entered the public markets on November 19, 2026, facing organized demonstrations in both New York and San Francisco. In New York, the group ACT UP NY gathered at 4 Times Square to protest the company's impact on the medical field. The group distributed flyers and held banners to highlight their opposition to AI in healthcare.

The San Francisco rally was organized by Ban RSI, a group focused on the technical risks of AI development. This event featured speakers including former AI researchers Jacob Coxon and Daniel Kokotajlo, as well as safety educator Rob Miles. These groups represent a coalition of concerns ranging from immediate social biases to long-term existential risks.

Beyond the physical protests, a joint petition from several organizations, including Action Center on Race and the Economy, called for a boycott of the offering. This campaign targets bank executives and institutional investors, urging them to stop using retirement savings or taxpayer money to fund AI infrastructure. They argue that these investments could lead to an AI bailout at the expense of public interests.

Risks in medical AI

A primary concern for the ACT UP NY group involves the integration of AI into hospitals and doctors' offices. They argue that these technologies could replace live medical screening or human staff, potentially altering patient outcomes. The group is particularly worried about how AI might influence specific treatments and the data used to train these systems.

The protesters also highlight the potential for AI to perpetuate systemic biases within the healthcare system. They specifically point to concerns regarding racial, gender, and anti-LGBTQ biases in automated medical tools. These issues are seen as immediate threats to equity in how different populations receive medical care.

The advocacy group notes that some of these practices are already being implemented in clinical settings. By protesting the IPO, they aim to force Anthropic to address these specific harms before the company scales its influence. They want to ensure that the pursuit of profit does not come at the cost of marginalized patients.

The recursive self-improvement debate

Ban RSI is specifically targeting a process they call recursive self-improvement, which occurs when an AI system performs the work of building more powerful AI. They want a prohibition on companies using systems more powerful than those available at the rally for research. Ideally, they are calling for an international treaty to govern these activities.

Anthropic’s prospectus takes a different view, describing the use of successive Claude versions to assist in building new models as a structural advantage. The company notes that these systems help with experiments, training-data generation, safety audits, and code review. They argue that while full autonomous self-improvement has not arrived, the current methods are necessary for development.

The company admits that a system autonomously designing its own successor could increase the risk of humans losing control. However, Anthropic states it would only pause development if other frontier developers did so verifiably. This creates a disagreement with protesters who want Anthropic to take unilateral, verifiable steps rather than waiting for the rest of the industry to act.

Governance and founder control

The financial structure of the IPO has drawn scrutiny regarding who will ultimately control the company's direction. The offering priced 980 million Class A shares at $83 each, totaling an $81.34 billion offering. Despite the public nature of the company, the voting power remains highly concentrated in the hands of the founders.

According to the prospectus, a single Class F share held by the founders' LLC carries 50.1% of the general voting power. Additionally, a Long-Term Benefit Trust generally holds the power to elect and remove the Class T directors. These directors make up the majority of the board, effectively linking the founders' vehicle to the company's high-level governance.

This structure creates a paired accountability question for investors and the public. Critics want to know if the company's development practices can be verifiably constrained by external oversight. They are questioning whether the current governance structure will actually uphold Anthropic's stated goal of developing AI for the long-term benefit of humanity.

What to watch

  • Whether Anthropic announces specific, verifiable steps to limit recursive self-improvement.
  • If the boycott of the IPO affects the participation of major university endowments.
  • How the company responds to specific demands regarding bias in healthcare AI.

Try it

Look up the Anthropic prospectus to see the specific breakdown of Class A, T, and F shares.

Sources and links

First reported by Unite.AI. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

More AI news