Most GenAI vendors are pushing vague, one-sided contracts with limited IP indemnity, no SLAs, data usage traps, and escalating lock-in. Procurement and legal teams are moving fast, but the contract language isn’t keeping up with the operational risk.
This white paper breaks down the 7 most dangerous clauses we’ve seen in enterprise GenAI agreements — with anonymised examples from real negotiations.
Vendors limit liability for AI-generated IP infringement, shifting all risk to the customer.
Broad data retention and training clauses that expose your proprietary information.
No uptime, latency, or accuracy guarantees — yet enterprise pricing.
Usage metering clauses that create future compliance and audit exposure.
Escalation mechanisms masked as “innovation” that compound costs as you scale.
Vendors disclaim responsibility for outputs, hallucinations, and downstream decisions.
Data portability restrictions and model dependencies that prevent switching vendors.
Whether you’re evaluating your first enterprise AI contract or renegotiating an existing one — this briefing gives you the contract intelligence to protect your organisation from GenAI’s hidden commercial risks.
Delivered to your inbox instantly. No sales call required.