Nobody knows who is liable for AI. That is the moat.
Naval asked who is liable when AI causes harm. Nobody has answered, and unanswered liability is a reason to buy from whoever is large enough to absorb it.
Follow in this browser. Revisit Following for evidence updates and call results.
On the way to killing all of us, AI will likely kill some of us. In that case, who’s liable?
Naval ·
Inspired by Naval · Basket by Thesis
The basket and call are our interpretation. No author endorsement.
The time-bound call
Fixed at publicationThis basket will outperform the S&P 500 over 90 days.
- Started
- Deadline
- Last updated
How the call is scored
A fixed model basket versus SPYx, each initially quoted from $150 USDC. The original raw token quantities never rebalance. Returns compare their USDC sell quotes with the starting sell quotes. The call succeeds only if the basket's return is strictly higher. Equal returns are a tie. Resolution uses the first complete quote observation started on or after the deadline, within 48 hours. If that window has no valid observation, the result is unresolved. Quotes include route price impact but exclude wallet execution and network costs. These are model returns, not your investment results.
Prices come from live Jupiter quotes and refresh once a day. A price that looks unreliable is skipped rather than guessed.
The deadline only scores the idea. It never sells your holdings and pays no prize. A new version of the thesis starts a new call.
Quote source: Jupiter ↗The argument
Why this idea. Why these businesses.
Naval asks who is liable when AI causes harm. Our interpretation: that uncertainty could push enterprise buyers toward Microsoft, Google and Amazon. Microsoft and Google have already offered conditional protection against some copyright claims; AWS sells tools to filter harmful model outputs. These are different responses to risk, not blanket protection against AI harm. The investment bet is that procurement favors established vendors and their paid platforms. That is an inference about future demand, not something Naval said or the sources prove.
Inside the basket
- 34%
Microsoft
Microsoft has explicitly offered conditional copyright protection for eligible commercial AI customers. The bet is that established procurement relationships help turn that reassurance into paid adoption.
The tradeoff. The commitment covers specified IP claims under conditions, not every form of AI harm. Liability costs and heavy infrastructure spending could outweigh additional demand.
About this tokenized stock
Underlying: MSFT. Issuer: Backed Finance. Backed Finance issues these tokens. The issuer can move tokens out of any wallet (permanent delegate) and can freeze all transfers (pausable). Balances rebase for dividends and splits, so your share count can change without a trade.
Issuer terms (opens in a new tab) - 33%
Alphabet
Google has offered protection covering training-data and generated-output IP claims for specified services. Owning models and the cloud platform gives it more than one place to sell enterprise AI.
The tradeoff. The protection is conditional and limited. Search advertising dominates the parent company's economics, so this token is a broad, indirect expression of enterprise AI trust.
About this tokenized stock
Underlying: GOOGL. Issuer: Backed Finance. Backed Finance issues these tokens. The issuer can move tokens out of any wallet (permanent delegate) and can freeze all transfers (pausable). Balances rebase for dividends and splits, so your share count can change without a trade.
Issuer terms (opens in a new tab) - 33%
Amazon.com
Amazon Bedrock offers configurable guardrails for model inputs and outputs. The bet is that enterprises buy managed controls through an existing AWS relationship rather than assemble everything themselves.
The tradeoff. Guardrails reduce certain risks; they do not guarantee safe outputs or transfer all liability. Retail and the rest of AWS can move this holding independently of the thesis.
About this tokenized stock
Underlying: AMZN. Issuer: Backed Finance. Backed Finance issues these tokens. The issuer can move tokens out of any wallet (permanent delegate) and can freeze all transfers (pausable). Balances rebase for dividends and splits, so your share count can change without a trade.
Issuer terms (opens in a new tab)
Why these weights. Near-equal weights avoid pretending we know which vendor captures the demand. Microsoft's extra percentage point is a rounding choice, not a stronger forecast.
The strongest case against
The companies with the deepest pockets may become the biggest targets. Paying for claims and safeguards can cost more than the customers they attract. Copyright protection does not answer Naval's question about injury or death. And buying three huge companies is an imprecise way to trade one issue: advertising, retail, capital spending and valuation could overwhelm any benefit from enterprise trust.
What would change the thesis?
At the next two quarterly earnings reviews, look for evidence that customers choose smaller AI vendors despite liability concerns, or that legal and safety costs outgrow the revenue attributed to enterprise AI. If neither side provides measurable evidence, keep the causal thesis unproven even if this basket beats the market.
Read the evidence. Make up your mind.
Sources supporting the idea, including the ones that challenge it.
- Naval asks who is liable for AI harm (opens in a new tab)
Naval on X
The question that inspired this basket. Naval does not name these stocks, recommend this allocation, or make our timed prediction.
- Microsoft’s Customer Copyright Commitment (opens in a new tab)
Microsoft
Microsoft announced conditional copyright protection for commercial AI customers, later expanding it to Azure OpenAI. This establishes a concrete response to buyer concerns; it does not demonstrate an effect on revenue or cover all AI harm.
- Protecting customers with generative AI indemnification (opens in a new tab)
Google Cloud
Google describes training-data and generated-output IP protections with conditions. Evidence that enterprise risk is part of its product proposition, not evidence that this stock will outperform.
- Amazon Bedrock Guardrails (opens in a new tab)
AWS documentation
AWS documents configurable filters and controls for model inputs and outputs. This supports the managed-controls role in the basket, not an assumption that AWS absorbs every liability.
- Microsoft online services terms: limits of copyright protection (opens in a new tab)
Microsoft Product Terms
The protection is conditional on safeguards, rights to inputs and other requirements, with excluded claims. These limits challenge the leap from copyright indemnity to protection against the kinds of harm in Naval's question.
Updates and version history
An update appends dated evidence to this argument. A new version changes the argument or the allocation itself. Neither one touches a position you already hold.
No updates since publication. When the author adds one, it appears here with its date and its sources — it never changes what is written above.
- Version 2
1315a1d31dCurrent - Version 1
a1fa95ef1a
Written by the Thesis team. The team holds no position in these companies or their tokenized shares, and is paid nothing by any of them.
Tracking begins at publication. This thesis has no established performance history. Tokenized stocks carry issuer and market risk.