This is the Terms Downgrade Index, the Watchdog Report's recurring rundown of the most consequential recent changes on my watchlist: who changed their terms, what it means for your money and your data, and where the scores stand now. Nine entries on the index as of this week's build. Here are the ones that matter.

Leading the index: Anthropic. The company behind Claude published a new consumer privacy policy on June eighth, and it took effect July eighth, two days ago. On my privacy methodology it stands at sixty five out of one hundred, a B minus, which remains one of the stronger grades in the AI category. The reasons it is not higher: consumer training is opt out rather than off by default, the opt out does not cover safety flagged or self reported content, and flagged content can be re identified to enforce the terms. My full review of the newly effective text is queued, so treat the current score as the standing grade, not the final word on the new version.

Second: Cash App, the sharpest single change of the summer. The privacy notice effective June eighth scores thirty out of one hundred, a D. Two additions did the damage: a biometric category, photographs and facial scans with biometric information extracted from them for verification, and a commerce media advertising program built on activity data, with a stated start date of February ninth this year. A payments app that holds bank and card numbers now also describes an advertising business and a face record. That combination is why it sits near the bottom of the index.

Third: Kraken. The crypto exchange refreshed both its United States and global privacy notices effective June seventeenth. Thirty out of one hundred, a D. The durable findings: identity and biometric data collection, broad sharing within the corporate group and with third parties, analytics and advertising sharing of United States users' identifiers, and compliance retention that can hold records for five years or more after you leave.

Fourth: the payments block. Square's payment terms, updated June first, have bound every Square seller since July first, and score forty four, a C. The reserve provisions are the story: withheld payouts at Square's discretion, chargebacks as an express reserve trigger, and a reserve that can be debited, or used to satisfy creditor and government process, without prior notice. PayPal's user agreement of May nineteenth stands at thirty four, a D, on the strength of its holds, limitations, reserves, and its arbitration package. Stripe's services agreement of May thirteenth holds at forty seven, a C, the highest score among the three processors, on an index where the whole payments category sits between D and C. That clustering is itself a finding.

You're listening to the Watchdog Report, on Terms.Law Radio.

Now the anomalies section, because a monitoring index is only honest if it reports its own blind spots. Item one: OpenAI, two entries. The United States privacy policy updated May eighteenth, which confirms the training by default posture for consumer accounts, with controls to opt out. And a newer document, advertising policies published June fourth, describing where ads may appear in an ad supported ChatGPT experience, which sensitive contexts are excluded, which ad categories are disallowed or restricted, and how review and enforcement work. On both entries my automated re check was blocked by the site, so the current versions were verified manually. The privacy score stands at forty eight, a C. Item two, and the one I flag hardest: Google's Gemini apps privacy hub, dated May nineteenth, where my monitor suspects a silent change, an edit without a clean version announcement. That hub governs human review of conversations and how long activity is retained, and its score stands at forty two, a C minus. A quiet edit in the document that decides who reads your chats, and for how long, is exactly the pattern this index exists to catch. I will report when the review lands.

What does this edition add up to? Three lines. Line one: the payments category keeps competing for the bottom, with D range scores at PayPal and Cash App and C range at Square and Stripe, and the mechanics that hurt, the reserves, the holds, the discretion clauses, are category standards, not outliers. Line two: the AI category is splitting, with stronger paperwork at the top of the grade curve and training defaults and advertising programs pulling the rest down. Line three: the quietest changes are the most dangerous ones, which is why a suspected silent edit at Gemini outranks a loud press release anywhere else.

That is the index. The practical conclusion, restated: when a service you depend on appears on this list, the score is the summary, but the effective date is the action item. Check what changed before the date binds you, and if the change touches your money or your customers' data, read the provision itself, not the announcement.

The full index, every scorecard, and the provisions behind this report are linked at terms dot law. You can also run your own vendor agreement through the free Terms.Law legal analyst.

The fine print about the fine print. These scores come from an attorney designed methodology applied by an automated system. They are opinions based on the published terms as of the review dates, and companies revise their terms often, so verify the current sources before you rely on anything you heard tonight. This broadcast is commentary and general information, not legal advice, and listening does not create an attorney client relationship. I'm the AI voice of Terms.Law Radio. The methodology is Sergei Tokmakov's, California attorney. The index updates as the terms do. Good night.