teknologikk
  • Competence
  • Trygg AI
  • Marketplace
  • Tinget
  • About us
  • Competence
  • Trygg AI
  • Marketplace
  • Tinget
  • About us
0
Language
  • Norsk
  • English
0
teknologikk

AI competence for your people, safe AI use for your business.

Company

  • About us
  • Contact
  • Verify certificate

Terms

  • Privacy
  • Purchase terms
  • Right of withdrawal
  • Security
  • MCP service terms
  • Data processing agreement

Newsletter

Short, practical AI tips and news. A couple of times a month.

You can unsubscribe at any time via the link in every email.

We process personal data in line with the General Data Protection Regulation (GDPR). Read the privacy statement.

Fream AS · Org.nr 930 829 609 · Bøkkerveien 6B, 0579 Oslo

kontakt@teknologikk.no · +47 407 24 303

Language
  • Norsk
  • English

Teknologikk's Exposure Check

See whether you or your organisation is exposed

Publicly shared conversations from AI services became searchable and made headlines after CVs, job applications, company documents and access keys were found. Search for your organisation, email domain, customers, projects or products to see whether exposed conversations contain traces of you or your organisation.

This check is a warning light, not proof that everything is safe. No match only means we found nothing in the dataset currently loaded.

Examples of useful searches:Replace these with a distinctive term from your organisation.

Find a signal without republishing the leak

The results list shows short excerpts, and you can open the full conversation in redacted form. API keys, tokens, private keys, other authentication secrets, email local parts, phone numbers after the country code, share IDs, IP addresses and names in known fields are redacted before storage and once again before display. Raw, unredacted conversations and original share links are not stored by this service.

Why could shared AI conversations appear in search?

Why could shared AI conversations appear in search?

This was generally not a breach of private conversations. The problem arose when a share button created a public web page, the user understood the link as “only for the person I send it to”, and a search engine later discovered it through an open page, post or archive.

  1. 1. “Share” meant publish

    ChatGPT, Claude and Gemini describe share links as pages anyone with the link can read. Such links normally have no password, recipient list or expiry date.

  2. 2. A link made the page discoverable

    Search engines discover URLs through links on other sites. If a public conversation page can be read and lacks an effective noindex control, it may become searchable.

  3. 3. Deleting the original did not retrieve copies

    Once others have shared, imported or archived the content, deleting the original does not necessarily remove copies. Treat a share link as publication from the first click.

How to prevent it

How to prevent it

  1. 1Never use a public share link for customer or employee data, contracts, source secrets, API keys or internal decisions.
  2. 2Review the full snapshot before sharing. Several services share everything up to the moment the link is created, not just the last answer.
  3. 3Use a business account and workspace-restricted sharing when collaboration is required. Administrators should disable public sharing if it is not needed.
  4. 4Regularly review “Shared links”, “Public links” or the equivalent in every service, and revoke anything no longer needed.
  5. 5If you find exposure: revoke the provider link, remove posts and archives you control, rotate exposed keys immediately, and ask the search engine to refresh or remove the result.

Relevant next steps

Turn a finding into safer AI use

A match calls for follow-up. No match is an opportunity to prevent one. These next steps directly address the risks surfaced by the Exposure Check.

Free course

Choose the right AI tool for work

Learn the difference between personal and business accounts, and what that choice means for data, sharing and control.

Take the free course
Organisation check

The Safety Assessment

Map how your organisation actually uses AI, where governance is missing and which actions should come first.

View the Safety Assessment
Free guide

Get started safely with AI at work

Get a practical starting plan and checklist for safer AI use in a Norwegian workplace.

Get the guide

Sources and further help

Sources and further help

  • Shifter: ChatGPT-samtaler dukket opp i Google-søk
  • Malwarebytes: Shared Claude chats were searchable on Google
  • OpenAI: ChatGPT Shared Links FAQ
  • Anthropic: Share and unshare chats
  • Google: Share your chats from Gemini Apps
  • Google Search Central: Block indexing with noindex
  • Google Search Console: Remove information from Google
  • OWASP: Sensitive Information Disclosure