An AI company going public may widen access to its financial rewards. People whose lives it changes still need influence that does not depend on owning shares.
On 26 October, Anthropic filed for a proposed public offering: https://www.anthropic.com/news/s1-sec. Its announcement left the price and number of shares undetermined. This is a step towards a stock-market listing, not a completed sale or evidence that AI's gains will be widely shared. It makes a larger question newly concrete: when more people can own part of an AI company, who gains authority over what it does?
My argument is that financial participation and public accountability must be built separately. A share can offer a claim on future rewards. It cannot, by itself, protect a worker's livelihood, a customer's independence or a citizen's ability to contest a harmful decision.
The proposed structure makes the distinction unusually visible. According to the filing (https://www.sec.gov/Archives/edgar/data/2130357/000162828026067168/anthropicpbc-sx1.htm), public Class A shares carry one vote each. A vehicle initially comprising seven co-founders would hold 50.1% of voting power on matters submitted to shareholders generally, including elections of directors outside the trust's remit. Separately, the Long-Term Benefit Trust generally has authority to appoint and remove directors constituting a board majority. Its special shares carry no economic rights. Associated Press reporting (https://apnews.com/article/ca2f4c282bef3201a392e59e3df2300c) also describes these safeguards and concentrations of power.
It would be wrong to flatten this into a story of founders controlling everything, or public investors having no votes. It would be equally wrong to treat the trust's public-benefit purpose as proof that everyone affected has an effective voice.
There is a serious case for the design. Research can take years; safety precautions can delay revenue. Leaders insulated from immediate market demands may be better able to refuse a profitable but dangerous use. Giving ordinary shareholders more control could strengthen commercial pressure rather than social protection. The useful question is how mission protection is tested and challenged when reasonable people disagree about the mission's meaning.
Consider a hypothetical Indian software-services employee. AI assistance might help her complete difficult work and acquire new responsibilities. Her employer might instead use the same tools to compress staffing. If she also owned a small investment in the technology supplier, a rising share price would not necessarily compensate for lost income. Her financial interest, her workplace interest and her interest as a citizen could pull in different directions.
The economic opportunity is real in principle. New capital can fund research, computing capacity and services that make expertise cheaper. Wider ownership can allow more savers to participate in success, while also exposing them to losses. But the amount people gain depends on how much they own. People without spare savings need benefits through better services, stronger earnings and lower costs. Ownership alone provides no mechanism ensuring those outcomes.
For an Indian business building on a foreign AI service, bargaining power may matter more than the supplier's shareholder count. Imagine a small accounting firm investing months in adapting its work around one provider. An abrupt price or service change could leave it with expensive retraining and migration. This is a possible dependency, not an allegation about Anthropic. India's Competition Commission, in its 2025 AI study release (https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2175607&lang=2®=48), recommended better access to infrastructure and capabilities to reduce entry barriers, alongside greater transparency.
Governance therefore needs routes that reach beyond investors. The G20/OECD corporate-governance principles (https://www.oecd.org/en/publications/g20-oecd-principles-of-corporate-governance-2023_ed750b30-en/full-report/component-9.html) recognise stakeholders' interests, information and participation, and call for affordable redress where rights are protected by law. For AI, applying that principle could mean independent complaint handling, protected channels for workers to report concerns, and customer contracts with usable exit arrangements. A remedy available only through expensive English-language legal correspondence would leave many people outside it.
Three futures are plausible. Patient capital and credible mission oversight could support useful tools while independent institutions give affected people recourse. An uneven future could spread investment opportunities while gains still favour large asset owners and customers with negotiating power. A harmful future could combine pressure to monetise with costly customer dependence and weak outside scrutiny. Neither a listing nor a trust determines which future arrives; their interaction with competition, contracts and public institutions matters.
Questions for founders and investors
Who can challenge a consequential decision without buying shares? What evidence would show that mission safeguards changed an outcome? Can a dependent customer leave at an affordable cost?
Founders should publish clear answers about authority and remedies. Investors should distinguish patient stewardship from insulation against legitimate challenge. Public buyers should negotiate continuity and exit terms before dependence grows. Governments should preserve enforceable protections for people with no financial stake. An AI company can invite the public into its capital structure; society still needs a say in the terms on which it enters people's lives.
Sources
Anthropic announcement, 26 October 2026: https://www.anthropic.com/news/s1-sec
SEC filing, 26 October 2026: https://www.sec.gov/Archives/edgar/data/2130357/000162828026067168/anthropicpbc-sx1.htm
Associated Press, 26 October 2026: https://apnews.com/article/ca2f4c282bef3201a392e59e3df2300c
G20/OECD principles, 2023: https://www.oecd.org/en/publications/g20-oecd-principles-of-corporate-governance-2023_ed750b30-en/full-report/component-9.html
CCI study release, 2025: https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2175607&lang=2®=48
On 26 October, Anthropic filed for a proposed public offering: https://www.anthropic.com/news/s1-sec. Its announcement left the price and number of shares undetermined. This is a step towards a stock-market listing, not a completed sale or evidence that AI's gains will be widely shared. It makes a larger question newly concrete: when more people can own part of an AI company, who gains authority over what it does?
My argument is that financial participation and public accountability must be built separately. A share can offer a claim on future rewards. It cannot, by itself, protect a worker's livelihood, a customer's independence or a citizen's ability to contest a harmful decision.
The proposed structure makes the distinction unusually visible. According to the filing (https://www.sec.gov/Archives/edgar/data/2130357/000162828026067168/anthropicpbc-sx1.htm), public Class A shares carry one vote each. A vehicle initially comprising seven co-founders would hold 50.1% of voting power on matters submitted to shareholders generally, including elections of directors outside the trust's remit. Separately, the Long-Term Benefit Trust generally has authority to appoint and remove directors constituting a board majority. Its special shares carry no economic rights. Associated Press reporting (https://apnews.com/article/ca2f4c282bef3201a392e59e3df2300c) also describes these safeguards and concentrations of power.
It would be wrong to flatten this into a story of founders controlling everything, or public investors having no votes. It would be equally wrong to treat the trust's public-benefit purpose as proof that everyone affected has an effective voice.
There is a serious case for the design. Research can take years; safety precautions can delay revenue. Leaders insulated from immediate market demands may be better able to refuse a profitable but dangerous use. Giving ordinary shareholders more control could strengthen commercial pressure rather than social protection. The useful question is how mission protection is tested and challenged when reasonable people disagree about the mission's meaning.
Consider a hypothetical Indian software-services employee. AI assistance might help her complete difficult work and acquire new responsibilities. Her employer might instead use the same tools to compress staffing. If she also owned a small investment in the technology supplier, a rising share price would not necessarily compensate for lost income. Her financial interest, her workplace interest and her interest as a citizen could pull in different directions.
The economic opportunity is real in principle. New capital can fund research, computing capacity and services that make expertise cheaper. Wider ownership can allow more savers to participate in success, while also exposing them to losses. But the amount people gain depends on how much they own. People without spare savings need benefits through better services, stronger earnings and lower costs. Ownership alone provides no mechanism ensuring those outcomes.
For an Indian business building on a foreign AI service, bargaining power may matter more than the supplier's shareholder count. Imagine a small accounting firm investing months in adapting its work around one provider. An abrupt price or service change could leave it with expensive retraining and migration. This is a possible dependency, not an allegation about Anthropic. India's Competition Commission, in its 2025 AI study release (https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2175607&lang=2®=48), recommended better access to infrastructure and capabilities to reduce entry barriers, alongside greater transparency.
Governance therefore needs routes that reach beyond investors. The G20/OECD corporate-governance principles (https://www.oecd.org/en/publications/g20-oecd-principles-of-corporate-governance-2023_ed750b30-en/full-report/component-9.html) recognise stakeholders' interests, information and participation, and call for affordable redress where rights are protected by law. For AI, applying that principle could mean independent complaint handling, protected channels for workers to report concerns, and customer contracts with usable exit arrangements. A remedy available only through expensive English-language legal correspondence would leave many people outside it.
Three futures are plausible. Patient capital and credible mission oversight could support useful tools while independent institutions give affected people recourse. An uneven future could spread investment opportunities while gains still favour large asset owners and customers with negotiating power. A harmful future could combine pressure to monetise with costly customer dependence and weak outside scrutiny. Neither a listing nor a trust determines which future arrives; their interaction with competition, contracts and public institutions matters.
Questions for founders and investors
Who can challenge a consequential decision without buying shares? What evidence would show that mission safeguards changed an outcome? Can a dependent customer leave at an affordable cost?
Founders should publish clear answers about authority and remedies. Investors should distinguish patient stewardship from insulation against legitimate challenge. Public buyers should negotiate continuity and exit terms before dependence grows. Governments should preserve enforceable protections for people with no financial stake. An AI company can invite the public into its capital structure; society still needs a say in the terms on which it enters people's lives.
Sources
Anthropic announcement, 26 October 2026: https://www.anthropic.com/news/s1-sec
SEC filing, 26 October 2026: https://www.sec.gov/Archives/edgar/data/2130357/000162828026067168/anthropicpbc-sx1.htm
Associated Press, 26 October 2026: https://apnews.com/article/ca2f4c282bef3201a392e59e3df2300c
G20/OECD principles, 2023: https://www.oecd.org/en/publications/g20-oecd-principles-of-corporate-governance-2023_ed750b30-en/full-report/component-9.html
CCI study release, 2025: https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2175607&lang=2®=48