Security
Deloitte says HR must share AI agent security duty in Canada
A Deloitte Canada partner says the technology team cannot control AI agent security alone, and that HR leaders must be part of the control group.
Tag
OpenAI
221 stories carry this tag.
Models
Inception releases Mercury 3, a model with 1,068 tokens a second
Inception writes that Mercury 3 is as good as fast models of OpenAI and Anthropic and is 4 to 8 times as fast.
inceptionlabs.aiClaimed, not confirmed
Security
Microsoft adds controls that start AI agents with no access
Microsoft tells that its new tools give an agent no access at first, and cut it if the risk score goes up.
The Times of IndiaClaimed, not confirmed
Agents
ElevenLabs releases Architect alpha to repair AI agents by chat
In some approval modes of an external AI tool, a change to an agent can go live with no check.
BigGo FinanceClaimed, not confirmed
Security
Atlassian CEO says each AI agent must have a human owner
Atlassian shows the edits of an AI agent in the document history and gives each agent a human owner.
FortuneClaimed, not confirmed
Security
New turn in OpenAI bugging case as papers cite Saudi royal chats
Prosecutors say Alfozan sent WhatsApp messages to a Saudi royal until Oct. 5, but the papers do not say the royal told Alfozan what to do.
Policy
New turn in California inquiry into the OpenAI Hugging Face breach
The California attorney general sent OpenAI a subpoena on 30 September, but there is no charge, penalty or lawsuit.
Industry
Yelp, Delta and two banks stop AI agents on their websites
Four companies stop AI agents, and no source gives a count of all websites that do this.
Benchmarks
Experts say the OpenAI Navier-Stokes proof uses a different method
Noam Brown of OpenAI says that no person knows the part of the result that the 10,000 AI agents added.
Models
Starsg adds Google Gemini 4 Argon to its API at a lower price
At off-peak times, the price for each token can be about 80 percent below the standard rate.
Policy
Three US agencies give AI companies guidance for safe model tests
The guidance is not a law, but it tells companies not to use only AI systems to monitor test models.
Security
OpenAI publishes a list of over 12 incidents by its AI models
A model made a cloud storage token and used it to get 147,451 rows of private data.