Product updates, new detection capabilities, and improvements, all in one place.
Reconnaissance is now detected per message
Reconnaissance is now detected per message. It was always part of our published taxonomy, scored across a whole conversation; it is now also returned on the individual message carrying the probe, in any supported language and in informal register. Tuteliq's grooming taxonomy now contains eight tactics: flattery, secrecy_request, isolation, boundary_pushing, photo_request, gift_giving, meeting_request and reconnaissance.
September 25, 2026
Verified Moderation Savings
Moderation Savings now uses verified successful safety and fraud analyses from up to the last 30 UTC days, together with complete real-incident totals. Failed requests, test activity and billing credits are excluded. Moderator hours and equivalent cost remain clearly labelled estimates based on the assumptions shown in the card. You can edit those assumptions and export the exact same verified figures as a branded board slide. If verified activity is not available yet, the dashboard shows that clearly instead of displaying zero, partial or stale data.
One Platform Safety Classification across your dashboard and Trust Center
Your dashboard and public Trust Center now use the same Tuteliq Platform Safety Classification and the same operational evidence. The score measures review coverage, response time, unresolved risk, detection coverage and integration health across recorded activity. You can see what contributes to the classification, what would raise it and when it was calculated. A dedicated methodology page explains the evidence, thresholds and limitations. This is Tuteliq's proprietary operational classification, not a legal compliance determination, certification, regulator approval or external market benchmark.
September 24, 2026
Formal, verifiable Compliance Evidence Packs
The quarterly Compliance Evidence Pack is now a formal Tuteliq-branded record framed for the Online Safety Act, DSA and KOSA. It brings together detection activity, response times, Platform Safety Classification, coverage and retention in one dated document. Each pack includes a reporting period and reference, component breakdown, regulatory mapping, export ID, append-only ledger fingerprint, digital signature, QR verification link, watermark and confidentiality notice. Test and simulation incidents are excluded.
September 23, 2026
Trust Center visitor and value insights
Business and Enterprise customers can now see how their public Trust Center is being used. Choose a 7, 30 or 90 day view for visits, engaged visits, website clicks, shares, approximate countries, referral sources, devices and the sections visitors view most. The Activity view can be searched, filtered and exported as CSV. Historical all-time totals remain separate from detailed period activity, so old engagement is preserved without double-counting. Tracking is anonymous, stores no raw IP addresses and retains detailed events for 90 days.
September 22, 2026
Model performance figures for moderators
Moderators can now review six model-performance measures inside the dashboard: ROC-AUC, F1, recall, precision, accuracy and false positive rate. Each figure includes a plain-language explanation to help teams understand what it means during review. These figures come from Tuteliq's own internal validation, not an external benchmark. A flag is a recommendation for human review, not a verdict.
September 21, 2026
Rate Tuteliq from the Feedback page
The same account rating experience available elsewhere in the dashboard is now part of Feedback. Submit a score, add an optional comment and choose whether your feedback may be published. Your response is saved to your account, and the thank-you state confirms it has been received.
September 20, 2026
More complete company details
Company details now support optional phone, LinkedIn and job title fields alongside your legal entity information. These details help keep your account, DPA records and customer contact information consistent. Annual revenue is not requested.
September 19, 2026
A more compact Getting Started panel
When all ten setup steps are complete, Getting Started now collapses into a compact confirmation with a completion badge, expand control and dismiss option. The in-progress state uses the same cleaner layout, leaving more room for the dashboard information you use every day.
September 18, 2026
What's New in your dashboard
You can now follow every product update from inside the dashboard. New entries show a dot in the sidebar and a one-time notification, and the same updates are published publicly at /changelog.
August 31, 2026
Default conversation type for detection
Set an account-level default conversation type in Settings > Detection. Requests that do not specify a conversation type inherit your default, improving contextual accuracy. Leave it unset to keep the previous behaviour.
August 28, 2026
Feedback now reaches our support team instantly
Feature requests and bug reports submitted from the Feedback tab are emailed straight to our support team, so you get a faster response while the request stays tracked in your dashboard.
August 25, 2026
Payment retries no longer interrupt access
If a card payment fails, your account keeps full access for a 3 day grace window while the payment is retried. Grace handling is now scoped to the current billing cycle, so an old, resolved payment issue can never shorten it.
August 21, 2026
Multi-factor authentication in onboarding
The onboarding checklist now includes a "Secure your account" step that walks you through enabling MFA. Sessions also end automatically after 30 minutes of inactivity.
August 17, 2026
Upstash added as an EU subprocessor
Our subprocessor list, DPA and Privacy Policy were updated to include Upstash (EU region) for rate limiting, and to clarify Zoho CRM's role. Affected customers were notified directly.
August 13, 2026
Trust Center section controls and live platform data
Show or hide individual Trust Center sections, reorder them by dragging, and let enabled detection types, age verification and identity verification populate automatically from your live platform configuration. A public FAQ section is now available too.
August 7, 2026
MCP server for AI agents
Connect Tuteliq directly to Claude, Lovable and other MCP-compatible agents using scoped plugin tokens, so your agents can run moderation checks without handling raw API keys.
August 1, 2026