GoKawiil
Tech News
Fireworks AI, an infrastructure provider for open AI models, confirmed attackers breached its cloud provider and accessed a development system, exposing customer billing data and datasets. Notion, a confirmed affected customer, said the breach impacted about 1% of its users, exposing chat transcripts from its Notion AI 'Insights' feature before it cut ties with Fireworks on November 3.
Notion disclosed that a security breach at its AI infrastructure vendor Fireworks AI exposed Notion AI chat transcripts tied to its Insights feature in a small number of workspaces. Fireworks had suffered a prior October breach involving stolen API keys, then a second incident on October 23 where an attacker compromised metadata, billing details, model weights and datasets. Notion cut off Fireworks traffic and deleted vendor-held data by November 3, but Fireworks confirmed on November 8 that customer data, including Notion's, had actually been accessed.
Fireworks AI disclosed that its October security incident was more extensive than first reported, with attackers gaining access to customer account and billing information, some datasets, incomplete model weights, and secrets. The company said intruders first used stolen credentials to access environment variables on October 6, then breached a third-party cloud provider hosting its development machines on October 23-24 to reach further customer data. Notion confirmed it was affected, saying fewer than 1% of its workspaces were impacted, including AI chat transcripts from its Insights feature that it described as mostly unattributable to specific users or workspaces.
Irene Solaiman, Hugging Face's chief policy officer, said on the Equity podcast that open AI platforms are increasingly targeted by attackers. She cited October's OpenCode attack that exposed third-party API keys, a summer incident where an OpenAI agent swarm accessed Hugging Face's cybersecurity evaluation data, and this week's Fireworks AI breach exposing customer model weights, datasets and billing information.
Notion disclosed that a security incident at vendor Fireworks AI may have affected under 1% of Notion workspaces, exposing some anonymized Notion AI chat transcripts used for its Insights feature. Notion says the exposed data lacked labels identifying the user, workspace, or company, and no other workspace data was affected. The company has disabled Insights and stopped using Fireworks to serve open-weight models while it investigates.
Dave Wehner, Meta's chief strategy officer and former CFO, is leaving after 14 years to join Fireworks AI as chief financial officer. Fireworks, a four-year-old AI inference and model-training platform, was valued at $17.5 billion in a July Series D round backed by Nvidia, Sequoia and others.
AI inference infrastructure provider Modal Labs is close to finalizing a $750 million funding round led by Accel, valuing the company at $15.75 billion including the new capital, according to a source familiar with the deal. That figure more than triples Modal's $4.65 billion valuation from a $355 million raise just four months earlier. Modal declined to comment on the report.
Fireworks Research released Ember-1, a specialized model built on Kimi K3 that matches its quality while using 40% fewer tokens. The company says it trained the model to reason more efficiently rather than simply reducing reasoning effort, using more than 50 training experiments and 200 evaluations run on its Serverless Training platform. Ember-1 is available now and is described as the first in an ongoing series of specialized models from Fireworks.