Skip to content

AI Flash · live feed

What is happening in AI right now

A timeline of everything that has passed through the radar — newest first, with an evidence status for each item.

2524 across the entire feed ↗ details, original sources and our own analyses

Continuously updated

Latest items

2524 items in the selected feed

How to read evidence statuses ↓

Friday, November 13

9 items
Security only one source so far

Research describes a vulnerability of vision-language models in image downscaling involving a hidden instruction

Research on arXiv describes a vulnerability called Scale-Induced Semantic Shift: a hidden instruction in an image is not triggered in vision-language models at 512×512, but activates after downscaling below 128×128. The ScaleGuard defense reduced the attack success rate from 61.4% to 24.3%.

arXiv cs.AI (Artificial Intelligence) Details and sources →

Thursday, November 12

15 items
Regulation and law only one source so far

Nvidia privately admitted the feasibility of tracking the location of AI chips

According to Congressman Bill Foster, Nvidia privately confirmed that it is technically capable of implementing the chip geotracking required by the Chip Security Act – despite the public position of the industry association SIA that such a measure is infeasible. The case is linked to an investigation into chip smuggling into China.

MIT Technology Review — AI section Details and sources →
Tools and apps only one source so far

AWS adds a guide for migrating knowledge bases to Amazon Bedrock Managed Knowledge Base to the Agent Toolkit for AWS

AWS has added a migration guide to the Agent Toolkit for AWS for moving a knowledge base from customer-managed to Amazon Bedrock Managed Knowledge Base. A coding agent proposes the migration plan; the managed version offers higher quotas, automatic management, and a different, usage-based pricing model.

AWS Machine Learning Blog Details and sources →