Two data breaches in two months is not a coincidence. It is a pattern. Fireworks AI has now exposed Notion customer chat data in what the company is calling its second security incident in as many months. For anyone who relies on AI-powered tools to manage workflows, this should raise a clear and urgent question: how much trust are you placing in the infrastructure behind your daily productivity? We think the answer demands more scrutiny than most users are currently giving it.
The timing makes this particularly uncomfortable. On one side of the ecosystem, companies like Siri co-founder backs Helix voice ID that runs on an iPhone 13 are proving that powerful AI can operate entirely on-device, keeping sensitive data local and out of third-party hands. On the other side, we see a remote access provider like N-able having to pull service after a zero-day disclosure, demonstrating how quickly trust evaporates when a vendor's perimeter is breached. Fireworks AI's repeated exposure sits squarely in the middle of these two realities: it is a cloud-based AI service that handles user data, and it has now proven twice that its security posture is not keeping pace with the sensitivity of the information it processes.
What does this mean for you in practical terms? If you use Notion's AI features or any service that routes your data through Fireworks, your chat histories and potentially your proprietary business conversations have been exposed, not once, but twice, in a very short window. This is not a hypothetical risk. It is a demonstrated failure. The pattern suggests that the company's response to the first breach did not resolve the underlying vulnerability, which is a much more serious concern than any single incident. For users who have been evaluating whether to deepen their reliance on AI-native tools, this is the kind of signal that should prompt a hard look at data residency, encryption practices, and whether the convenience of cloud-based AI is worth the exposure.
The specific consequence to watch is how Notion and other Fireworks customers respond. If they stay quiet and continue using the service without demanding a transparent post-mortem, they signal that repeated breaches are acceptable. If they push for auditable security standards or begin migrating to on-device alternatives, like the voice ID system that runs entirely on an iPhone 13, they set a new expectation for the entire AI services market. Your takeaway is straightforward: do not wait for a third breach to evaluate where your data lives and who can access it. The technology to keep your information local already exists. The question is whether you are ready to use it.