Future of Life Institute

Future of Life Institute (FLI) conducts advocacy, research and grantmaking concerning risks from transformative technologies. Its AI-policy position emphasizes large-scale harm, loss of human control and concentration of power. It advocates public oversight and restrictions where it considers development risks unacceptable.

Organization focus: Transformative technology and large-scale risks.

Founded: 2014, according to its financial history.

Policy activity: Research, public advocacy and proposed governance measures.

Official site: futureoflife.org

Last source review: September 12, 2026

The 2023 pause proposal

FLI's March 2023 open letter called for a public, verifiable pause of at least six months in training systems more powerful than GPT-4. It proposed that governments impose a moratorium if laboratories could not arrange the pause promptly. The scope was a category of advanced training activity, rather than all AI research or every existing application.

The letter called for independent scrutiny and shared safety protocols during the pause. Its accompanying governance agenda included oversight of powerful systems and computing resources, auditing, liability, provenance mechanisms and public support for safety research. These were demands addressed to laboratories and policymakers, not evidence that a binding international pause took effect. The original letter and later FLI positions also need to be read separately: a proposal tied to a particular model and date is not necessarily the organization's entire subsequent program.

Proposed public institutions and obligations

Policymaking in the Pause, published in April 2023, expanded the letter into specific policy recommendations. It proposed independent auditing and certification for defined systems, regulation of access to computing resources, capable national AI agencies, liability for harm, measures addressing model leaks, safety-research funding and standards for synthetic content. The paper treats organizational capacity and enforcement as necessary companions to technical testing.

Its proposed auditing regime would examine risks before deployment and require disclosure of residual risks. Its recommendations also contemplated agencies able to set reporting duties and act on findings from impact assessments. The paper discussed the EU AI Act while it was still a proposal; those historical descriptions should not be substituted for the enacted instrument's current requirements. The document establishes what FLI advocated in 2023, not a current legal obligation on developers.

Current published position and its limits

The position page reviewed on September 12, 2026 supports a pause on frontier experiments and a moratorium of at least 15 years on developing artificial superintelligence. It connects an end to restrictions with safety guarantees, broad benefit and global deliberation. The page also opposes extreme corporate and governmental concentration of power and favors applications intended to assist people.

Although the page displays a May 2024 publication date, its currently displayed wording is reported here as observed in September 2026; that date is not evidence that every present sentence appeared in the original version. Its predictions about controllability and risk are FLI's assessments. The position describes a desired governance settlement, and does not establish scientific consensus or the legal status of any particular AI system.

Funding and decision-making

FLI's finances page identifies early support from Elon Musk for AI-safety grantmaking after its 2014 founding and a major unconditional contribution from Vitalik Buterin in 2021, part of which functions as an endowment. Its 2025 account says that spending was largely supported by those endowment-like funds. The page separately reports grants, staffing, professional services, outreach and other expenditure categories.

FLI states that its board and internal teams determine policy positions. Its conflict-of-interest description identifies Jaan Tallinn as a donor who has also served on the board, says Buterin has no decision-making role, and states that it does not accept donations from Big Tech or companies seeking to build AGI. These are disclosed arrangements and institutional commitments; they do not make every donor a sponsor of every FLI proposal.

Selected policy documents

• Pause Giant AI Experiments (March 2023).
• Policymaking in the Pause (April 2023).
• Our Position on AI, current text reviewed September 2026.

Related articles

• United States policy on catastrophic AI risk
• European Union policy on catastrophic AI risk
• EU AI Act
• Centre for the Governance of AI