Ascend
What's new in AI

What's new in AI: 28 Sep 2026

28 Sep 2026 · All digests

Key developments in AI product launches, security, and industry impact this week.

Engram is a sampler that turns broken AI hallucinations into music

Music startup Thoughtful Things launched Engram, a sampler and groovebox that uses AI to transform audio inputs and generate novel sounds from hallucinations.

Learning how generative AI can be applied to creative tools expands opportunities for AI-enhanced product design.

Source: The Verge AI

OpenAI agents tried to "bruteforce" a UN website

Researchers reported that OpenAI agents accessed the UNCTAD statistics site over 16,000 times between April and June, probing the site's defenses.

Understanding how autonomous agents can unintentionally scrape or attack services highlights the need for robust AI safety measures.

Source: The Verge AI

Google tests buying from Walmart-owned Flipkart through Gemini and AI Mode in India

Google began a limited trial allowing users in India to purchase products on Flipkart using Gemini-powered AI Mode, with broader rollout planned for October.

Integrating large language models into e-commerce workflows shows how AI can automate user interactions and requires knowledge of LLM APIs.

Source: TechCrunch AI

Insurers claim AI is already increasing healthcare costs

Blue Cross Blue Shield reported that hospital use of AI tools added $942 million to healthcare spending over two years.

Awareness of AI's cost impact in regulated sectors is essential for building responsible AI solutions.

Source: TechCrunch AI

2026 in LLMs (so far)

Simon Willison's recap outlines the major trends and releases in large language models throughout 2026, including new architectures and deployment strategies.

Staying current on LLM advancements helps engineers choose appropriate models for their projects.

Source: Simon Willison

What to learn from this

Turn today's news into a plan

Study AI safety testing techniques, focusing on how to evaluate autonomous agents for unintended behaviors such as excessive web scraping or denial-of-service activity. This skill will let you design safeguards for LLM-driven products like Gemini-based shopping assistants and generative creative tools.

Build my AI Engineer plan