Today, The Washington Post covered critical vulnerabilities depthfirst found in TikTok. These vulnerabilities allowed hackers to access anything on a user’s device that TikTok itself could access, including the camera, microphone, payment information, photos, and the user’s entire TikTok account. Read more in the thread 🧵

We disclosed the vulnerabilities to TikTok and worked with them to patch it before making this public. AI has changed cybersecurity. Even the most secure, scrutinized, and well-funded companies, trusted by billions of users with sensitive personal information, can be breached. Security experts know this. But despite the attention AI security is receiving, consumers don't always understand how directly these risks can affect them today, or have taken precautions. We are sharing this news to show them that the risks are real and that they should be more cautious with the level of access they give to digital services. The more capabilities and autonomy we give AI, the more we need to invest in the security that makes it possible. @depthfirstlabs exists for this reason. We found these vulnerabilities using our platform and dfs-large1, our flagship AI cybersecurity model post-trained on GLM 5.2 through reinforcement learning.

@washingtonpost @depthfirstlabs this is incredibly important work to raise awareness

@washingtonpost @depthfirstlabs 🤯

@washingtonpost @depthfirstlabs this is not critical rank #5 for today below openai monorepo hack by 3 indian guy

@washingtonpost @depthfirstlabs Weird.

@IceSolst @washingtonpost @depthfirstlabs Ya I’m uninstalling TikTok