This Team Is Rethinking AI’s Core: Perforated AI Bets On Dendrites...

Perforated AI adds dendrite-like units to standard neural networks so each neuron can do more. In tests, their Perforated Backpropagation cut compute and model size dramatically (examples: ~38x cost reduction and 158x CPU speedups) while keeping or improving accuracy. If it scales, teams could train and run models on local hardware instead of renting huge GPU clusters.

ORIGINAL LINK: https://www.forbes.com/sites/ronschmelzer/2025/08/29/this-team-is-rethinking-ais-core-perforated-ai-bets-on-dendrites/

Anthropic Reaches Settlement in Landmark AI Copyright Case with US Authors...

Anthropic settled a lawsuit with U.S. authors who claimed the company used pirated books to train its AI without permission. The case avoided a big trial and could affect how AI companies handle copyrighted work in the future. Details of the settlement will be shared soon, but no money amount was revealed.

ORIGINAL LINK: https://petapixel.com/2025/08/29/anthropic-reaches-settlement-in-landmark-ai-copyright-case-with-us-authors/

Apple employees built an LLM that taught itself to produce good user interface code - but worryingly, it did so independently...

Apple researchers taught a model to write working SwiftUI code even though the training data had almost no Swift examples. They ran cycles of generation, compilation checks, and GPT‑4V filtering to produce nearly one million valid programs and a stronger model called UICoder. The result shows self‑improvement can work but raises worries about reliability and scaling of synthetic datasets.

ORIGINAL LINK: https://www.techradar.com/pro/apple-employees-produced-an-llm-that-taught-itself-to-produce-good-user-interface-code-independently

ChatGPT conversations are monitored and can be reported to police, OpenAI confirms...

OpenAI monitors ChatGPT conversations and can report some to police. Special systems and human reviewers flag threats. Self-harm cases are not reported and the model gives supportive responses.

ORIGINAL LINK: https://www.indy100.com/science-tech/chatgpt-conversations-reported-to-police

Which AI search should you trust for facts? Our librarians’ best rankings. - The Washington Post...

We tested nine AI tools with hard questions to see if they give correct answers. Librarians helped judge the results and compare them to Google searches. Some AI did well, but others gave worse answers than Google.

ORIGINAL LINK: https://www.washingtonpost.com/technology/2025/08/27/ai-search-best-answers-facts/

When Crisis Meets AI...

Crisis-driven AI adoption taught people to use AI fast but not to judge it. Now anyone can engineer "seemingly conscious" AIs that act human and trick people, causing harm. We must quickly build deception literacy, rules, and social norms to spot and resist this threat.

ORIGINAL LINK: https://aiforbeginners.substack.com/p/when-crisis-meets-ai

Why AI gets stuck in infinite loops — but conscious minds don’t...

Seth says living minds are deeply embedded in time and driven by entropy. That time‑anchoring helps animals avoid infinite loops and adapt in open‑ended worlds. Classical AI, built on timeless computation, likely cannot achieve the same conscious, time‑based intelligence.

ORIGINAL LINK: https://bigthink.com/neuropsych/anil-seth-consciousness-time-perception/