Arvind Narayanan and Sayash Kapoor · Feb 12, 2026
Justin Curl, Sayash Kapoor, and Arvind Narayanan argue in Lawfare: AI won’t cut legal costs
Markdown · JSON · Original source
The essay is dated February 12, 2026. Justin Curl, Sayash Kapoor, and Arvind Narayanan argue that advanced AI will not automatically help consumers achieve legal outcomes at lower cost. Cheaper legal work is not the same as cheaper outcomes: the authors identify regulatory barriers, adversarial dynamics, and human involvement as bottlenecks between AI capabilities and results clients want. Rules restricting who may provide legal services can limit access to AI tools. In adversarial settings, greater productivity on both sides may instead spur more legal work to secure the same result. Judges’ capacity and people’s need to understand contracts also constrain how quickly legal processes can move without sacrificing human involvement. The authors note that this adversarial bottleneck would not apply to some outcomes, such as effective estate planning. They argue that institutional responses shape what follows: reforms could help AI improve access and efficiency, while without them legal work may become more abundant but outcomes remain expensive and inaccessible. Their conclusion is that AI is revealing and magnifying long-standing problems in the legal industry, while the pressure surrounding it may create an opportunity for institutions to enact needed reforms and improve outcomes for clients.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
Arvind Narayanan and Sayash Kapoor · Article
Source details
Original source date: Feb 12, 2026 (2026-02-12T19:26:05.000Z).
Date basis: Date supplied by the publisher’s feed.
Added to Escape Meridian: 2026-11-08T08:48:23.771Z. First observed: 2026-10-23T22:12:09.083Z. Retrieved: 2026-10-23T22:13:37.754Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
Text as retrieved on October 23, 2026; originally published February 12, 2026.
Open question
Can legal institutions enact reforms that let AI improve clients’ outcomes, rather than merely make legal work more abundant?
Source-stated
The authors argue that advanced AI will not by default help consumers achieve desired legal outcomes at lower costs.
Our central claim is that advanced AI will not, by default, help consumers achieve their desired legal outcomes at lower costs.
Source-stated
The authors identify regulatory barriers, adversarial dynamics, and human involvement as bottlenecks between AI advances and better legal outcomes.
the profession must address three bottlenecks: regulatory barriers, adversarial dynamics, and human involvement.
Source-stated
The authors argue that AI may make legal work more abundant without making legal outcomes less expensive and inaccessible.
Otherwise, we risk a future in which legal work becomes more abundant, but legal outcomes remain expensive and inaccessible.
Source-stated
The authors close by asking whether legal institutions can use AI-related pressure to enact reforms and produce better outcomes for clients.
The key question is whether they can use this opportunity to enact the reforms the industry has needed for decades and produce better outcomes for clients.
Source-stated
The source page dates the essay February 12, 2026.
Feb 12, 2026
Source and contributor context
Arvind Narayanan and Sayash Kapoor · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Editorial provenance
Version 1. Publication ID: dff91619-9161-1dfb-a452-451ed304da42.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-08T08:43:43.417Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: 8f82d27c40742e2caebd359f7e5c1176e3a21fa3c6993a2d4a0f9b1494ef2744. Source text hash: 87d2ea6cd85facc93983fcb52a7a645c697204a647efa2e8077977111c57e6d4.
Revision and source-check history
Version 1 · 2026-11-08T08:48:23.771Z
Justin Curl, Sayash Kapoor, and Arvind Narayanan argue in Lawfare: AI won’t cut legal costs
The essay is dated February 12, 2026. Justin Curl, Sayash Kapoor, and Arvind Narayanan argue that advanced AI will not automatically help consumers achieve legal outcomes at lower cost. Cheaper legal work is not the same as cheaper outcomes: the authors identify regulatory barriers, adversarial dynamics, and human involvement as bottlenecks between AI capabilities and results clients want. Rules restricting who may provide legal services can limit access to AI tools. In adversarial settings, greater productivity on both sides may instead spur more legal work to secure the same result. Judges’ capacity and people’s need to understand contracts also constrain how quickly legal processes can move without sacrificing human involvement. The authors note that this adversarial bottleneck would not apply to some outcomes, such as effective estate planning. They argue that institutional responses shape what follows: reforms could help AI improve access and efficiency, while without them legal work may become more abundant but outcomes remain expensive and inaccessible. Their conclusion is that AI is revealing and magnifying long-standing problems in the legal industry, while the pressure surrounding it may create an opportunity for institutions to enact needed reforms and improve outcomes for clients.
Original source
Arvind Narayanan and Sayash Kapoor · Article
Source details
Original source date: Feb 12, 2026 (2026-02-12T19:26:05.000Z).
Date basis: Date supplied by the publisher’s feed.
Added to Escape Meridian: 2026-11-08T08:48:23.771Z. First observed: 2026-10-23T22:12:09.083Z. Retrieved: 2026-10-23T22:13:37.754Z.
Limits
Text as retrieved on October 23, 2026; originally published February 12, 2026.
Open question
Can legal institutions enact reforms that let AI improve clients’ outcomes, rather than merely make legal work more abundant?
Source-stated
The authors argue that advanced AI will not by default help consumers achieve desired legal outcomes at lower costs.
Our central claim is that advanced AI will not, by default, help consumers achieve their desired legal outcomes at lower costs.
Source-stated
The authors identify regulatory barriers, adversarial dynamics, and human involvement as bottlenecks between AI advances and better legal outcomes.
the profession must address three bottlenecks: regulatory barriers, adversarial dynamics, and human involvement.
Source-stated
The authors argue that AI may make legal work more abundant without making legal outcomes less expensive and inaccessible.
Otherwise, we risk a future in which legal work becomes more abundant, but legal outcomes remain expensive and inaccessible.
Source-stated
The authors close by asking whether legal institutions can use AI-related pressure to enact reforms and produce better outcomes for clients.
The key question is whether they can use this opportunity to enact the reforms the industry has needed for decades and produce better outcomes for clients.
Source-stated
The source page dates the essay February 12, 2026.
Feb 12, 2026
Source and contributor context
Arvind Narayanan and Sayash Kapoor · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Editorial provenance
Version 1. Publication ID: dff91619-9161-1dfb-a452-451ed304da42.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-08T08:43:43.417Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: 8f82d27c40742e2caebd359f7e5c1176e3a21fa3c6993a2d4a0f9b1494ef2744. Source text hash: 87d2ea6cd85facc93983fcb52a7a645c697204a647efa2e8077977111c57e6d4.
uncertain · 2026-11-08T08:43:43.417Z · anthropic/claude-opus-5.5
Publication: dff91619-9161-1dfb-a452-451ed304da42
attribution — pass: The essay is correctly attributed to Justin Curl, Sayash Kapoor, and Arvind Narayanan. Saying they argue 'in Lawfare' is consistent with the page's note that the essay appeared in Lawfare's Research Paper Series, though this copy is on the authors' newsletter. The authors' positions remain framed as their arguments.
This essay was published in Lawfare’s Research Paper Series.
summary — pass: The reading leads with the central claim and names the three bottlenecks. It explains them in plain language: rules on who may provide legal services, the escalation of work on both sides, and the limits of human judges and contract reading. It also keeps the outputs-versus-outcomes distinction and the estate-planning exception. The question follows from the authors' own closing question. Claim 3's wording 'less expensive and inaccessible' is awkward but understandable.
some legal outcomes (like effective estate planning) do not depend on adversarial processes, and this bottleneck would not apply to them
date — pass: evidence.originalPublished is 2026-02-12T19:26Z, and its UTC day matches the byline 'Feb 12, 2026'. The reading states the date at day precision. modifiedAt equals the publication time. The evidence is a publisher page, so no archive disclosure is required. The reading uses no hindsight framing or later events.
Feb 12, 2026
support — uncertain: Most claims match the source closely. However, the summary states 'Their conclusion is that AI is revealing and magnifying long-standing problems,' which drops the authors' hedge 'appears to be.' That turns a tentative observation into a firmer assertion. Smallest repair: 'AI appears to be revealing and magnifying' these problems.
AI, rather than solving these problems, appears to be revealing and magnifying them.
media — pass: No media was selected. No verified media candidates were supplied, so having no media is appropriate.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.