Anthropic opens bug bounty program to everyone, promising rewards of up to $10,000.



Anthropic, known as the developer of the 'Claude' series of AI models, has opened its bug bounty program to the public. Previously, only security researchers could report bugs, but now anyone can participate in the program, report bugs, and receive rewards.



Anthropic's bug bounty program has traditionally been a closed program for the security researcher community, and the product has been enhanced by numerous discoveries.

Starting May 8, 2026, this program became publicly available on the security platform HackerOne, and any registered user can now participate.

Anthropic | Bug Bounty Program Policy | HackerOne
https://hackerone.com/anthropic?type=team



Rewards are categorized based on whether they are 'core' or not, and on four levels of severity. The details are as follows:

low Medium high emergency
non-core part $100 - $250
(Approximately 15,700 yen to 39,200 yen)		 $250 to $750
(Approximately 39,200 yen to 118,000 yen)		 $1,000 to $2,500
(Approximately 157,000 yen to approximately 392,000 yen)		 $3,000 to $5,000
(Approximately 470,000 yen to 784,000 yen)
core part $250 to $750
(Approximately 39,200 yen to 118,000 yen)		 $1,000 to $2,500
(Approximately 157,000 yen to approximately 392,000 yen)		 $3,000 to $5,000
(Approximately 470,000 yen to 784,000 yen)		 $7,500 to $10,000
(Approximately 1.18 million to 1.57 million yen)


As of the time of writing, the reporting rates were 12.15% for 'low,' 43.32% for 'medium,' 43.72% for 'high,' and 0.81% for 'urgent.'

The program's total payouts are $550,835 (approximately 86.3 million yen), with 429 cases reported and $199,450 (approximately 31.3 million yen) paid out in the last 90 days.

Related Posts:

in AI, Posted by logc_nt