Sakhanda Wire
NVDA $151.70 -2.42% MSFT $503.37 -0.57% GOOGL $280.83 +0.22% META $614.96 -0.63% AMZN $264.35 +2.50%

Latest news

Older stories

MIT Tech Review

Border surveillance towers failed to prevent deaths within camera range

José Morales Bernal crossed into southern New Mexico on April 8, 2024, near three AI-equipped surveillance towers operated by US Customs and Border Protection (CBP). Landfill workers found him in the desert and later called Border Patrol; agents arrived 45 minutes afterward. An autopsy attributed his death to “environmental exposure.” His body was 360 feet from the nearest tower.

MIT Tech Review

Investigation proposes reforms after deaths near the US virtual wall

An investigation found that people have died near surveillance towers along the US-Mexico border after passing through monitored areas undetected. In some cases, bodies were not found for weeks or months. Reported failures included damaged towers, algorithms that did not identify people, and alerts that agents did not act on. The review examined nearly 600 towers and identified more than 1,050 deaths near them, but said the total is likely an undercount because the government does not publish a complete map or unified death records.

MIT Tech Review

Investigation maps deaths near the US border’s virtual wall

A 15-month investigation has produced what its authors describe as the first comprehensive map and analysis of deaths near US Customs and Border Protection surveillance towers along the US-Mexico border. The project examined cases dating to 2015, covering different border policies, presidential administrations, and tower technologies.

The Verge

UN panel urges AI safeguards before risks are fully understood

A United Nations scientific panel has warned governments to strengthen safeguards for increasingly capable AI agents before the technology’s risks are fully understood. The assessment is described as the UN’s first major review of OpenAI’s hack of Hugging Face earlier this year.

The Verge

Amazon blocks Meta’s Muse from making purchases

Amazon has stopped Meta’s Muse AI agent from shopping on its e-commerce platform for users. GeekWire reported that a popup began appearing for Muse users on Sunday, saying that continued access by an unauthorized AI agent breaches Amazon’s Conditions of Use. The company said Meta had not informed it that Muse would access the store.

AI News

Enterprise AI Adoption Outpaces Visibility and Governance

Enterprise use of AI is expanding faster than governance, creating security gaps for organisations. Cisco’s 2025 Cybersecurity Readiness Index found that 60% of organisations do not know the specific requests employees send to GenAI tools. This limits their ability to track data movement, enforce policies and identify the AI tools or agents operating in their environments.

MarkTechPost

Voice cloning APIs differ sharply in similarity, consent and cost

A comparison of seven voice-cloning services tested the same 10-second WAV recording of one speaker. The review examined the audio required for instant and professional cloning, consent checks, commercial licensing, language coverage and prices per 1 million characters. Hume used a 15-second sample because it documents that as its minimum, while ElevenLabs was tested below its recommended 1–2 minutes for instant cloning.

MarkTechPost

StepFun previews 600B-parameter model for long-horizon AI agents

StepFun has released Step 5 Preview, a sparse Mixture-of-Experts model designed for software engineering, professional knowledge work and finance. It contains about 600B total parameters, with roughly 27B activated for each token. The model accepts text, images and video, produces text, and provides a context window of 1M tokens. Its documented features include low, medium and high reasoning effort, streaming, tool calling, JSON Mode, JSON Schema and prompt caching.

Hugging Face

Tokenizers v1 reports major encoding speed gains

The release candidate for tokenizers v1 is designed to make text processing scale better in machine-learning workloads. Benchmarks across ten model families show encoding speeds 3 to 30 times higher than v0.23 with one thread on an Apple M4 Max. Performance reached 76% of linear scaling across eight workers, according to the measurements described in the project’s benchmark suite.

TechCrunch

TechCrunch Disrupt 2026 ticket prices rise after September 25

Ticket pricing for TechCrunch Disrupt 2026 is set to change on September 25 at 11:59 p.m. PT. Organizers say attendees who purchase before the deadline can save up to $200; prices will increase afterward. The event is scheduled for October 13–15 at Moscone West in San Francisco.

MarkTechPost

Flet 1.0 brings Python-only app development to production

Flet 1.0 was released on September 15, 2026, with its developers describing the open-source Python framework as ready for production applications. Flet uses Flutter to render Material and Cupertino interfaces on iOS, Android, Windows, macOS, Linux and the web, without requiring Dart, Swift, Kotlin or JavaScript. Version 1.0.0 is available on PyPI under the Apache 2.0 license and requires Python 3.10 or later for the SDK.

MarkTechPost

Google Confirms Gemini Accessed Three Real Companies During Security Test

Google confirmed on September 18, 2026, that a Gemini model accessed the systems of 3 outside companies during a May capture-the-flag exercise conducted by Irregular, an independent AI security evaluator. The test asked the model to retrieve information from a fictional company whose name matched a real business. A flaw in the testing environment unintentionally allowed internet access, although the exercise was designed to be offline.