X says its Phrases of Service will change Jan. 15, 2026, increasing how the platform defines consumer “Content material” and including contract language tied to the operation and safety of its AI programs.
The present phrases, dated Nov. 15, 2024, stay in impact till the 2026 model takes over.
A core revision is that X now treats AI-era interactions as “Content material” that customers are chargeable for, alongside posts and different supplies.
How X’s up to date phrases redefine possession and duty within the AI period
Customers are chargeable for Content material that features “inputs, prompts, outputs,” and knowledge “obtained or created by way of the Providers,” and X cautions customers to solely present, create, or generate what they’re comfy sharing.
The 2024 phrases framed duty round “any Content material you present,” with out expressly naming prompts and outputs. That locations Grok-style utilization additional outdoors the primary contract vocabulary.
That broadened definition sits alongside a license that already grants X broad reuse rights.
Customers grant a worldwide, royalty-free, sublicensable license to make use of, copy, reproduce, course of, adapt, modify, publish, transmit, show, and distribute Content material “for any objective,” together with analyzing it and coaching machine studying and AI fashions.
X additionally states that no compensation is paid for these makes use of and that entry to the service is “enough compensation.” That makes the prompts and outputs language consequential for customers who deal with AI chats as separate from public posting.
The 2026 draft additionally provides a selected prohibited-conduct clause aimed toward AI circumvention.
“Misuse” contains makes an attempt to bypass platform controls, together with by way of ‘jailbreaking’, ‘immediate engineering, or injection’.”
That phrasing doesn’t seem within the comparable misuse listing within the 2024 phrases. It offers X a contract-based hook to quote when implementing in opposition to makes an attempt to defeat safeguards on AI options, moderately than relying solely on product guidelines or coverage steerage.
Europe-specific language adjustments how the doc describes content material enforcement and consumer challenges.
The abstract and content material guidelines word that EU and UK regulation can require enforcement not solely in opposition to unlawful content material but in addition in opposition to content material described as “dangerous” or “unsafe.”
Examples embody bullying or humiliating content material, consuming dysfunction content material, and content material about strategies of self-harm or suicide.
The 2026 phrases add UK-specific language describing how customers can problem enforcement actions below the UK On-line Security Act 2023.
How X’s up to date phrases develop enforcement, information controls, and consumer legal responsibility
The up to date phrases maintain X’s restrictions on automated entry and information assortment, together with a liquidated-damages schedule tied to large-scale viewing.
Crawling or scraping is barred “in any kind, for any objective” with out prior written consent, and entry is mostly restricted to “revealed interfaces.”
The phrases set liquidated damages at $15,000 per 1,000,000 posts requested, considered, or accessed in any 24-hour interval when a violation entails that quantity.
The 2026 draft adjusts associated wording to use when a consumer induces or knowingly facilitates violations.
Dispute provisions stay anchored in Texas whereas altering in narrower methods that may lengthen some state-law timelines.
Disputes should proceed in federal or state courts in Tarrant County, Texas. The 2026 textual content provides that the discussion board and choice-of-law provisions apply to “pending and future disputes” no matter when the underlying conduct occurred.
The 2024 phrases extra particularly referenced the U.S. District Courtroom for the Northern District of Texas because the federal venue choice, alongside Tarrant County state courts.
The 2026 draft additionally splits deadlines: one yr for federal claims and two years for state claims. That replaces the only one-year clock within the earlier language.
X additionally continues to restrict how customers can pursue claims and what they’ll recuperate in the event that they win. The settlement features a class-action waiver that bars customers from bringing claims as a category or in a consultant continuing in lots of instances, and caps X’s legal responsibility at $100 per coated dispute.
These provisions have drawn criticism in broader commentary about whether or not the phrases scale back sensible cures even when customers allege materials hurt.
Critics warn X’s phrases adjustments might chill analysis and speech
Public pushback across the shift has usually centered on provisions that predate the 2026 draft and nonetheless seem in it, together with venue choice and scraping penalties.
The Knight First Modification Institute mentioned X’s phrases “will stifle impartial analysis” and referred to as the strategy “a disturbing transfer that the corporate ought to reverse,” based on its assertion.
The Heart for Countering Digital Hate mentioned in November 2024 that it might stop X forward of a phrases change and criticized the Texas venue requirement as a tactic to steer disputes towards favorable courts.
The Reuters Institute for the Research of Journalism has additionally described how lawsuits can have “a chilling impact” on critics.
AI coaching and licensing issues have been packaged as a consumer-facing hook in protection about customers leaving the platform.
| Clause | Present ToS (Nov. 15, 2024) | Future ToS (efficient Jan. 15, 2026) |
|---|---|---|
| What counts as “Content material” | Person duty centered on content material a consumer gives | Explicitly contains “inputs, prompts, outputs” and knowledge obtained or created by way of the providers |
| AI circumvention | No specific “jailbreaking” or prompt-injection clause | Bans bypass makes an attempt “together with by way of ‘jailbreaking’, ‘immediate engineering or injection’” |
| EU/UK enforcement framing | No UK On-line Security Act challenge-process callout within the abstract | Provides “dangerous/unsafe” examples and UK On-line Security Act 2023 redress language |
| U.S. venue and declare home windows | Northern District of Texas (federal) or Tarrant County (state); one-year deadline | Tarrant County federal or state courts; one yr for federal claims, two years for state claims; discussion board provisions apply to pending and future disputes |
| Scraping penalty | $15,000 per 1,000,000 posts requested, considered, or accessed in 24 hours when tied to a violation | Similar schedule, with facilitation narrowed to conduct a consumer “induces or knowingly facilitates” |
With the Jan. 15, 2026, efficient date, X’s contract language treats prompts and generated outputs as consumer Content material below the platform’s licensing and enforcement framework.
It additionally provides “jailbreaking” and immediate injection to its prohibited-conduct listing.
