We Read the Terms of 175 AI Tools Twice. 60% Changed in 15 Weeks.

The numbers

First pass, 5 September 2026: 81 of 175 tools had changed since we first verified them on 23 June, 74 days earlier. That is 46%.

Second pass, 3 October 2026: 56 of 175 tools had changed again in just 28 days. That is 32% in four weeks.

Combined: 105 of 175 tools, 60%, changed at least once in 15 weeks. 32 tools changed in both passes. Only 52 were reachable both times and unchanged both times.

Direction matters more than volume. Across both passes, 31 scored factors moved toward more risk for the creator and 6 moved toward less. That is roughly five to one against you. 15 tools tightened commercial rights on their free tier, and 41 claims that a free tier was safe to monetize stopped being supported by the tool's own current wording.

We do not hold accounts at these tools, so we cannot say how many emailed their users. What we can say is that we found every one of these changes the only way a creator reliably can: by re-reading the page.

Five changes that would have cost a creator money

Lexica. In June its own terms supported commercial use of free images. On 3 October the terms read: "You must have a paid Lexica plan to use Lexica images commercially." Free use is now defined as personal use, "usage of Lexica images within a project that is not monetized in any way." A channel built on free Lexica images is now outside its licence.

PicLumen. Its pricing page used to say that all images and videos generated with PicLumen can be used for commercial purposes. The free Basic plan now reads "No commercial use rights", and ownership is "subject to the commercial license terms of your subscription plan."

Suno. A new Terms of Service took effect on 3 September 2026. The free plan on the pricing page now lists "No monthly song downloads" and "No commercial rights".

LOVO and PlayHT. Both are gone. LOVO, Inc. filed for Chapter 7 liquidation in May 2026 and every lovo.ai address now returns a suspended-deployment error. play.ht no longer resolves at all. If you used either for monetized work, the licence records you would point to in a dispute are no longer online.

And one change in the creator's favour, which shows why re-reading matters in both directions. Google rewrote its Flow help page and removed the per-tier visible "made with Veo" watermark. Visible marking is now a user toggle, applied automatically only for residents of India, South Korea and Vietnam. For weeks our own page was still telling readers to buy a $100 to $200 a month plan to escape a watermark they could switch off for free. We fixed it. It is the clearest example we have of a stale verdict costing a reader real money.

The mistake we found in our own data

Re-reading the terms also meant re-reading our own work, and it turned up a systematic error. In 18 places, a factor we had scored for a tool's free tier was backed by a quote that actually described a paid plan. A sentence beginning "If you are on the Creator plan" or a feature row headed "Everything in Free, plus" was being used as evidence about the free tier. Every one of those quotes was verbatim and from the tool's own page, which is why no check caught it. The words were right. Their scope was wrong.

We downgraded every affected factor to "unclear" rather than guessing, and added an automated check that now blocks our site from publishing if a free-tier safety claim rests on paid-plan wording. The October pass found two more cases with different phrasing, and those are fixed too. We are publishing this because a site that asks you to trust its quotes should show you what happens when its own quotes are wrong.

What this means if you monetize AI output

Read the licence on the day you generate, and keep a copy. A screenshot of the terms page with the date in the URL bar costs nothing and is the only record that survives a tool rewriting or deleting its page.

Treat a free tier as a demo unless the tool's own terms grant commercial use in words you can quote. Marketing copy, FAQ answers and feature tables change faster than contracts, and in several cases this year they contradicted the contract outright.

Re-check before you scale. A tool that was safe when you started a channel may not be safe when the channel starts earning. Over these 15 weeks, the odds that a given tool changed something were better than even.

Every tool page on ClipJury shows the date its evidence was last read, and links to the exact source. Our full list of changes is on the terms-change log.

How we did it

For each of the 175 tools, we stored every factor in our Commercial Monetization Risk score alongside the exact quote that justifies it, the URL, the source type and the date. In each pass we fetched every one of those URLs again and checked whether the quote still appears, whether the URL still resolves, and whether the licence, pricing, free tier or watermark had changed. Pages that block automated reading were opened in a real browser where possible. Where we still could not read a page, we say so on the tool's page rather than assume nothing changed. Only a tool's own terms, licence, pricing page or official statement can support a finding that its free tier is safe.

Tools mentioned

Lexica

Safe plan: Subscribe to the Starter (individual) plan; once you generate/download on a paid plan, those images are licensed for commercial use forever even after cancelling.

PicLumen

Safe plan: Only monetize images you generated yourself; do not directly reuse other users' public creations for commercial work, PicLumen's FAQ limits those to reference only.

Suno

Safe plan: Pro plan, $8/mo billed yearly ($96/yr), grants commercial-use rights to new songs.

Veo (Google)

Safe plan: Free tier, with the Visible watermark toggle off (verify in your region)

LOVO (Genny)

Safe plan: Basic, $24/mo

PlayHT

Safe plan: Log in and read the current plan comparison and Terms of Service in-app (or via docs.play.ht/legal) to confirm commercial license and any watermark before monetizing free-tier audio.

FAQ