Contract Negotiation

State of play.

• The Musk v. OpenAI trial is live and producing contract-law precedent in real time. Greg Brockman's personal diary has emerged as central evidence on the nonprofit-to-for-profit conversion, with testimony cutting against Musk's deception narrative — and the case is now the highest-visibility test of founder-agreement enforceability and fiduciary duties owed to departed board members in AI ventures (→ Brockman's Diary Revealed in Musk-OpenAI Trial First Week).
• AI vendor pricing is shifting from per-seat to consumption-based models, with Salesforce, Workday, and OpenAI all moving toward "agentic work units" or token-based billing — a structural change that makes AI costs variable rather than fixed and leaves measurement methodologies largely undefined in current contract terms .
• Elite firms are bypassing legal-tech middlemen and contracting directly with AI labs. Freshfields has deployed Claude firmwide through a direct Anthropic partnership and is building proprietary agentic workflows, pressuring legal-tech vendors to differentiate beyond base-model access .
• Supply chain disputes are escalating to litigation as courts tighten force majeure and performance defenses, tariff-driven cost pressure intensifies, and manufacturers respond by renegotiating or nearshoring — with 72% of supply chain professionals citing U.S. tariff changes as their top concern, per a 2026 industry survey .
• For counsel advising enterprise clients on AI procurement, supply agreements, or technology vendor relationships, the practical baseline is: consumption-based pricing terms need measurement definitions and cost caps before deployment; AI service agreements need performance warranties and documented renewal communications; and supply contracts need flexible performance triggers and contingency clauses that courts will actually enforce.

Where things stand.

• Founder-agreement and fiduciary-duty doctrine for AI ventures is being tested at trial. The Musk v. OpenAI trial centers on whether the 2015 nonprofit founding documents constitute enforceable contracts and what fiduciary duties Musk retained after leaving the board in 2018 — questions with no settled precedent in the AI context (→ Brockman's Diary Revealed in Musk-OpenAI Trial First Week).
• AI contract tooling is now embedded in standard workflows. Microsoft's Legal Agent, released in Word via the Frontier program, performs clause-by-clause review against customizable playbooks and generates redlines with tracked changes — using deterministic rather than generative AI, with attorney verification required . AI-powered platforms from Clio, Ironclad, Bind, and GC.ai are reducing contract review cycles from 30-90 minutes per round to seconds, with firms reporting 30-50% faster negotiations overall, per National Law Review reporting .
• AI service agreement disputes are reaching federal court. A California law firm's suit against Connex alleges that AI system malfunctions justified non-renewal and that the vendor's response — harassment and litigation threats — created independent liability, surfacing the reliability gap between AI demonstrations and production performance as a live litigation issue .
• Supply chain contract boilerplate is failing in court. Courts are rejecting loose force majeure and performance defenses as tariff pressure, freight volatility, and capacity constraints persist — with 57% of firms renegotiating contracts and 51% nearshoring, per 2026 industry data .
• The Pentagon-Anthropic blacklist remains in effect for defense contractors. The D.C. Circuit denied Anthropic's emergency stay; existing Claude deployments in military work must cease and new Pentagon contracts cannot incorporate the system, while a conflicting San Francisco district court ruling leaves the scope across non-Pentagon federal agencies unsettled .
• Non-monetary settlement terms are underweighted in mediation strategy. Analysis from Miles Mediation documents a recurring pattern: parties resolve financial terms before negotiating non-disparagement, reinstatement, policy change, and confidentiality provisions — the terms that most affect whether resolutions hold, particularly in identity-based employment disputes .
• AI disclosure in negotiations is unsettled. Whether parties must affirmatively disclose their use of AI tools in contract negotiations — and what protocols govern that disclosure — remains an open question as adoption accelerates .

Latest developments.

• Musk v. OpenAI trial underway: Brockman diary testimony, Musk settlement threats, and 2017 equity dispute testimony all in evidence (→ Brockman's Diary Revealed in Musk-OpenAI Trial First Week)
• Major AI vendors shifting to consumption-based pricing — Salesforce "agentic work units," Workday "units of work," OpenAI token-based billing — with measurement terms undefined in most current contracts
• Freshfields Lab deploys Claude firmwide via direct Anthropic partnership; Google Gemini already deployed to 5,000 professionals; tech-agnostic, non-exclusive structure designed to avoid vendor lock-in
• California law firm sues Connex in federal court over AI phone system malfunctions and alleged coercive contract retention tactics
• Supply chain disputes escalating to litigation in beauty, fashion, and automotive sectors; courts tightening force majeure defenses; 10% ad valorem tariff effective February 24, 2026, raising effective rates on Chinese goods to 22-34%
• Miles Mediation publishes analysis on non-monetary settlement terms and sequencing errors in mediation strategy
• Freshfields-Anthropic multi-year partnership announced; Claude usage surged 500% within six weeks of deployment
• Microsoft Legal Agent released in Word via Frontier program — clause review, redlines, version comparison, deterministic resolution
• National Law Review publishes guidance on managing AI-assisted counterparties in contract negotiations; disclosure norms remain unsettled

Active questions and open splits.

• Founder-agreement enforceability in nonprofit-to-for-profit AI conversions. Musk v. OpenAI is the first major trial on whether early-stage AI founding documents constitute binding contracts and what fiduciary obligations survive board departure — the outcome will set the reference point for structuring AI venture governance documents (→ Brockman's Diary Revealed in Musk-OpenAI Trial First Week).
• Consumption-based AI pricing — what goes in the contract. Vendors are billing on "agentic work units" and tokens with measurement methodologies that remain undefined; no market-standard contract terms have emerged for rate structures, audit rights, cost caps, or dispute resolution over billable-unit calculations .
• AI service agreement performance warranties and renewal coercion. The Connex case surfaces the reliability gap between AI demonstrations and production performance as a live litigation issue — and whether vendor conduct during non-renewal constitutes actionable coercion is now before a federal court .
• AI disclosure obligations in negotiations. Whether parties must affirmatively disclose AI tool use in contract negotiations — and what ethical or contractual consequences attach to non-disclosure — is unsettled as adoption accelerates across platforms .
• Force majeure and performance defense viability in tariff-disrupted supply chains. Courts are tightening the standard as tariff pressure persists; the question is how specifically performance triggers, transparency provisions, and contingency clauses must be drafted to survive judicial scrutiny .
• Pentagon blacklist scope across federal agencies. The D.C. Circuit and the San Francisco district court have issued conflicting rulings on whether the Anthropic designation extends beyond Pentagon contracts — leaving defense contractors and other federal agencies in an uncertain compliance posture pending the May 19 oral arguments .
• Non-monetary settlement sequencing. Whether the documented pattern of resolving financial terms before non-monetary provisions constitutes a negotiating error that counsel can be faulted for — and how to structure mediation agendas to avoid it — is an open tactical and professional-responsibility question .

What to watch.

• Verdict or significant rulings in Musk v. OpenAI — particularly on founder-agreement enforceability and the fiduciary-duty standard for departed board members in AI ventures (→ Brockman's Diary Revealed in Musk-OpenAI Trial First Week).
• May 19 D.C. Circuit oral arguments in the Anthropic blacklist case — outcome will determine compliance obligations for defense contractors and potentially reshape federal AI procurement policy .
• Whether market-standard contract terms emerge for consumption-based AI pricing — measurement methodologies, audit rights, and cost caps are the critical gaps .
• Whether the Connex federal suit produces rulings on AI performance warranty standards or vendor coercion liability — either outcome will influence how AI service agreements are drafted .
• Whether additional BigLaw firms announce direct AI-lab partnerships on the Freshfields model, accelerating pressure on legal-tech vendors to differentiate .
• Whether supply chain litigation produces published decisions tightening the force majeure standard in tariff-disrupted contexts — the drafting implications for pending contract renegotiations are immediate .