BTC $69,016.12 -6.35%
ETH $1,947.19 -5.11%
BNB $635.80 -6.18%
XRP $1.00 -3.80%
SOL $78.75 -5.37%
TRX $0.3116 -1.67%
DOGE $0.0638 -5.18%
ADA $0.1685 -2.64%
BCH $217.57 -5.34%
LINK $9.76 -3.65%
HYPE $71.23 -1.13%
AAVE $111.42 -6.30%
SUI $0.7772 -4.99%
XLM $0.1429 -2.38%
ZEC $934.85 -4.99%
AAPL $356.43 +0.10%
AMZN $268.00 +0.70%
GOOGL $300.10 +1.90%
MSFT $521.12 -0.11%
META $625.77 -0.38%
NVDA $164.00 +2.03%
TSLA $398.10 +2.86%
SNDK $927.24 -2.93%
INTC $60.73 +0.10%
SPCX $140.46 +0.75%
MU $794.00 +0.42%
AMD $574.09 +2.22%
BTC $69,016.12 -6.35%
ETH $1,947.19 -5.11%
BNB $635.80 -6.18%
XRP $1.00 -3.80%
SOL $78.75 -5.37%
TRX $0.3116 -1.67%
DOGE $0.0638 -5.18%
ADA $0.1685 -2.64%
BCH $217.57 -5.34%
LINK $9.76 -3.65%
HYPE $71.23 -1.13%
AAVE $111.42 -6.30%
SUI $0.7772 -4.99%
XLM $0.1429 -2.38%
ZEC $934.85 -4.99%
AAPL $356.43 +0.10%
AMZN $268.00 +0.70%
GOOGL $300.10 +1.90%
MSFT $521.12 -0.11%
META $625.77 -0.38%
NVDA $164.00 +2.03%
TSLA $398.10 +2.86%
SNDK $927.24 -2.93%
INTC $60.73 +0.10%
SPCX $140.46 +0.75%
MU $794.00 +0.42%
AMD $574.09 +2.22%

now

All
Article
Flash

first_img Knowledge base application Notion reports less than 1% workspace AI conversation leaks

The knowledge management application Notion has released an announcement acknowledging that some users' AI chat records have been leaked. The company stated that the affected workspaces account for less than 1% of all workspaces and is currently notifying the affected users. If users have provided their account, password, or credentials to the AI, Notion recommends treating them as compromised, immediately checking the login records of the relevant accounts, and changing passwords or credentials.The source of this incident is a hacker attack on the service provider Fireworks. Notion uses Fireworks AI to process data for a feature called Insights, which helps workspace administrators understand how team members are using AI. To generate this analytical data, Notion sends the saved AI conversation content to Fireworks for processing. Fireworks has also served as the inference service provider for some open-weight models in Notion's products.Notion stated that the potentially leaked data mainly consists of user chat records, including the chat content input by users and the replies given by the AI. The announcement explained that some users may have sent their account, password, or various credentials to the model while using AI. Notion advises the relevant users to immediately change these passwords or credentials and check for any unusual activity in their account login records.

Anthropic applies zero-knowledge proof technology to AI reasoning to prevent model swapping

Anthropic has open-sourced FlashProve, bringing the commonly used zkSNARK (zero-knowledge proof) technology from the cryptocurrency field into AI model inference.This technology was originally designed to prove the validity of blockchain transactions without disclosing transaction details. Anthropic aims to use it to solve another problem: when users pay a high price for an AI model's API, how can they know if the service provider has secretly switched to a cheaper model?FlashProve allows service providers to first generate encrypted fingerprints for the model weights and then provide mathematical proof for the inference results. Users do not need to obtain the model weights or rerun the model; they can check whether the specified output comes from the previously committed set of weights. Claude was also involved in the development and optimization of this system.Anthropic has already completed experiments on Meta's 400 billion parameter model Llama 4 Maverick. For every 1024 tokens of context processed, the generated proof is about 19 MB, and verification on a regular CPU takes about 17 seconds.However, currently, each proof can only verify one output token and cannot validate an entire response. The GPU computation time required to generate the proof is also about 70 to 95 times that of regular inference. Whether it can be used to identify API service providers that secretly switch models in the future will depend on when the cost of proof generation can be reduced.
app_icon
ChainCatcher Building the Web3 world with innovations.