If you’ve ever shared a Claude conversation using a public link, it might be worth checking your privacy settings. Over the weekend, users discovered that publicly shared Claude chats and Artifacts were appearing in Google Search and other search engines through queries such as site:claude.ai/share. The indexed pages reportedly included everything from resumes and business documents to API keys, health information, and other sensitive conversations that users likely never expected to become searchable.
Importantly, this wasn’t the result of a hack or data breach. Instead, the conversations were shared using Claude’s built-in Share feature.
Claude allows users to generate a public share link, creating a snapshot of a conversation that anyone with the URL can access. That snapshot also includes any Artifacts such as interactive documents, apps, or code generated during the conversation. While the feature is intended for collaboration, users discovered that some of these public pages had also been indexed by search engines, making them discoverable through simple Google searches.
UPDATE, ITS WORSE THAN THE CHATS
the same thing is happening with shared artifacts. every app, doc, dashboard and tool people published from claude is also sitting indexed and searchable
people have already pulled up internal company dashboards, full project plans with client… https://t.co/ZzezhlDbom pic.twitter.com/Ljk3zzXL4N
The issue first gained traction after Reddit users noticed hundreds of searchable Claude conversations over the weekend. Reports suggest that many users assumed “anyone with the link” meant the content wouldn’t appear in search results unless someone explicitly shared the URL with them, leading to understandable confusion once Google indexing came to light.
Anthropic says the behavior wasn’t the result of a bug or data breach, but rather how its sharing feature is designed to work. The company told TechCrunch that Claude chats remain private by default, and it doesn’t publish chat directories or sitemaps for search engines. However, if a user shares a public Claude link on a website, forum, or social platform that can be crawled, search engines may index it like any other publicly accessible webpage.
The company’s Privacy Center also states that shared chats create a public snapshot of the conversation, including any Artifacts generated before the link was created. Users can revoke access at any time by heading to Settings > Privacy > Shared Chats and unsharing the conversation. That said, previously indexed pages may continue to appear in search results or cached archives until they’re refreshed.
Google, meanwhile, says the responsibility lies with the website hosting the content. In a statement to TechCrunch, spokesperson Ned Adriance said search engines don’t decide which pages are made public, noting that Google provides website owners with controls to prevent crawling or indexing and respects those directives. The incident serves as a timely reminder that once content is made publicly accessible on the web, it can often be treated like any other webpage — even if it started as a private AI conversation.
AI companies spend enormous amounts of time convincing us their technology is safe and secure. But what happens when those same companies become targets of increasingly sophisticated AI-powered attacks? Anthropic, the company behind Claude, is now facing that question. According to a Washington Post report, cybersecurity startup Cathedral discovered vulnerabilities that let its researchers access sensitive information in Anthropic's internal systems.
The researchers used AI agents to identify potential entry points in less than 12 hours. Interestingly, they didn't need to crack Claude itself. Instead, they exploited weaknesses in third-party business applications that Anthropic uses for everyday operations. Anthropic's security testing program disclosed the findings, which allows outside researchers to identify and report vulnerabilities.
Military working dogs charge into places the rest of us would flee. So, I understand why anyone would want to breed tougher, healthier dogs for the job.
Warhound, a new startup from Colossal Biosciences co-founder Ben Lamm and Anduril co-founder Palmer Luckey, has stepped out of stealth with $200 million to clone and gene-edit working dogs for American troops, police departments, and emergency services (via TechCrunch).
AI can describe a picture, identify objects, and answer questions about what a camera sees. But when a blind user needs a quick answer, getting an entire spoken essay about the scene isn't exactly helpful. Researchers at the University of Maryland (UMD) have found that AI-powered visual assistance can leave blind users doing too much work to get the information they actually need.
The study, titled Say It My Way: Exploring Control in Conversational Visual Question Answering with Blind Users, examined how people interact with Be My AI, an AI-powered feature of the Be My Eyes app. The researchers found that users often received lengthy descriptions that didn't adequately address their immediate goals, forcing them to ask follow-up questions or find ways to steer the AI towards a more useful answer.