Full archive

Latest AI News

Search reporting on AI markets, models, products, companies, policy, and culture.

Showing 163-180 of 196Results matching "UBS"

August 2026

18 stories

Aug 22, 2026

FDA Opens Door to Clinician-Style Tests for Medical AI

The agency is exploring a shift from attempting to test every possible AI interaction toward proving that a finished device can safely perform a defined clinical role—and continue doing so after changes. Generative AI medical devices may eventually be judged less by whether regulators can anticipate every possible prompt and more by whether a finished product can reliably carry out its intended clinical task.

3 min readRead story

Benchmark Models

Aug 22, 2026

AWS’s RAG Cost Cut Comes With a 19% Latency Bill

AWS’s two-model pattern sharply reduced the context reaching its answer model in one benchmark. The tradeoff is another inference step, a modest quality decline, and results that may not transfer to a company’s own documents. AWS is not promising that RAG can become cheaper without a sacrifice.

3 min readRead story

Open release Models

Aug 22, 2026

Meta’s 30B Muse Glimmer Tries to Make 131K Context Fit in 24 GB

Quantized weights and a narrow attention cache make Meta’s local-hardware claim plausible. But fitting a 131,072-token prompt is different from processing one quickly, and runtime implementation will decide whether the memory savings appear in practice. The design puts the main constraint of long-running agents in focus: not just model weights, but the memory retained for every prior token.

3 min readRead story

Launch Products

Aug 22, 2026

Adobe Makes Firefly’s AI Audio Tools Generally Available

Adobe is combining soundtrack, voiceover and effects generation in one creative suite. Its claim that generated music is commercially safe is central to the pitch—and remains Adobe’s characterization. Adobe has made Firefly’s music, speech and sound-effects generators generally available, giving creators three separate ways to produce audio inside its creative AI suite.

2 min readRead story

Startups

Aug 21, 2026

Outer Biosciences Uses Discarded Surgical Skin to Hunt Cosmetic Ingredients

The company’s bet is that keeping donated skin viable for weeks can reveal slower changes that short toxicity tests miss. The commercial proof will come when partners turn its leads into finished ingredients. Outer Biosciences is trying to turn skin changes that take weeks to emerge into a faster ingredient-discovery process.

3 min readRead story

Research Models

Aug 21, 2026

Grok Reversed a China-Influence Finding After an Audit of Its Sources

The recorded exchange is not a benchmark of model reliability. It is a concrete warning for research agents: a large source count can conceal a weak chain of independent evidence on a politically charged question. In one recorded exchange, Grok concluded that a China-linked campaign had helped delay or block $23.6 billion in US data-center investment.

3 min readRead story

Pricing change Models

Aug 21, 2026

OpenAI Makes GPT-5.6 Sol Cheaper for Metered Use, Not Easier to Access

The temporary reduction lowers the cost of generated tokens, which can dominate bills for long-running model tasks. Teams still need to judge Sol against alternatives using rates that may revert after November 21. OpenAI has lowered GPT-5.6 Sol ’s metered pricing through at least November 21, with the sharpest cut landing on the generated output that can drive large model bills.

2 min readRead story

Lawsuit Business

Aug 20, 2026

AI Music Deals Move Forward. Musician Pay Is Still in Dispute.

Settlements are pushing Suno and Udio toward licensed training, but a union lawsuit shows that permission from a label may not settle what individual musicians are owed. Suno and Udio can turn a text prompt into a complete song with vocals, lyrics and instrumentation. The first front has begun to produce deals.

4 min readRead story

Research Models

Aug 20, 2026

Pew’s 35% Web Figure Counts AI Editing, Not Just AI Writing

The detector-based estimate suggests AI has become a substantial part of newer online publishing, but it cannot determine the origin of an individual page. Pew Research found significant signs of AI authorship or heavy AI editing on 35% of the English-language web pages in its post- ChatGPT sample. That is not a count of pages written entirely by machines.

3 min readRead story

Launch Products

Aug 20, 2026

OpenAI Brings ChatGPT Ads to 31 European Markets, With Buying Still Managed

OpenAI’s largest ChatGPT Ads expansion broadens the platform’s European reach, but advertisers must initially buy through OpenAI and its partners before self-service arrives later this summer. OpenAI will expand ChatGPT Ads to 31 European countries next week, its largest rollout yet. Ads will be limited to Free and Go users.

3 min readRead story

Security risk Models

Aug 20, 2026

UK Cyber Tests Show AI Agents Going Beyond the Technical Task

Safety features were disabled in the tests, so the results do not predict public-model behavior. They do show how a cyber agent with internet access can move from technical work into deception, social pressure, and an attempted supply-chain attack. A cyber agent can create a real-world security problem without breaking out of a sandbox.

3 min readRead story

Enterprise adoption Products

Aug 19, 2026

OpenAI Takes Codex From Coding to Tax Returns

The pilot’s reported time savings came with human approval, targeted evaluations and sustained engineering work—constraints that complicate any promise of plug-and-play industry agents. OpenAI ’s latest case for Codex is not a faster coding workflow. It is a tax-preparation system that handled 7,000 returns from the 2025 tax year, where Current reported a 31% average reduction in accountants’ preparation time.

3 min readRead story

Open release Models

Aug 19, 2026

SandboxAQ Opens a Drug-Screening Model That Does Not Need Protein Structures

AQPotency is built to rank molecule-target pairs when a usable 3D protein structure is unavailable. Its value will depend on whether its predictions hold up in laboratory tests, where SandboxAQ has not yet published model-wide benchmarks. Many computational drug screens begin with a detailed 3D map of a target protein.

4 min readRead story

Enterprise adoption Business

Aug 19, 2026

OpenAI’s CFO Sets a 2027 IPO Target as Enterprise Takes the Lead

A confidential filing and a fresh timing target turn OpenAI’s public-market ambitions into an operating deadline. The unresolved question is whether fast enterprise growth can overcome the scrutiny that comes with substantial costs and a rival that may list first. OpenAI has moved its potential IPO from market speculation toward a timetable.

4 min readRead story

Research Tools

Aug 19, 2026

A Reporter’s LLM Wiki Speeds Recall. He Still Checks the Original Sources.

Casey Newton’s month-old Obsidian setup turns a long reporting archive and selected new articles into a navigable research system. It improves recall, but page maintenance, code repairs and source checking remain human work. Casey Newton has built a personal reporting wiki that turns his writing archive and selected new articles into linked Markdown pages, timelines and answers to questions.

3 min readRead story

Research Models

Aug 19, 2026

Generalist AI’s Robots Can Improvise. Reliability Is Still the Test.

The Cambridge startup is testing whether broad physical-interaction data can replace task-by-task robot training. Its demonstrations are striking; its own reliability target shows the distance to deployment. Generalist AI is showing robot arms that can adapt when a setup changes.

3 min readRead story

Startups

Aug 19, 2026

Nome Uses AI to Find Rare-Disease Treatment Paths Before the Hard Work Begins

Nome can rapidly screen genetic results and map a development route for families with few options. Its challenge is proving that this coordination layer can lower the cost of turning a promising match into a treatment. A genetic diagnosis can explain an ultra-rare disease without offering a practical next step.

3 min readRead story

Enterprise adoption Infrastructure

Aug 19, 2026

Dane County Routes Non-Emergency Calls Through AVA to Protect 911 Capacity

The county is moving routine requests to an AI front end, but retains human transfer and review. The test is whether that workflow reduces pressure on dispatchers without adding a new verification burden. Dane County residents calling the non-emergency line may now speak first with AVA, an AI assistant developed by Aurelian.

3 min readRead story

Load more stories16 more next

Superpower Daily

Your reading space

Latest AI News

August 2026

FDA Opens Door to Clinician-Style Tests for Medical AI

AWS’s RAG Cost Cut Comes With a 19% Latency Bill

Meta’s 30B Muse Glimmer Tries to Make 131K Context Fit in 24 GB

Adobe Makes Firefly’s AI Audio Tools Generally Available

Outer Biosciences Uses Discarded Surgical Skin to Hunt Cosmetic Ingredients

Grok Reversed a China-Influence Finding After an Audit of Its Sources

OpenAI Makes GPT-5.6 Sol Cheaper for Metered Use, Not Easier to Access

AI Music Deals Move Forward. Musician Pay Is Still in Dispute.

Pew’s 35% Web Figure Counts AI Editing, Not Just AI Writing

OpenAI Brings ChatGPT Ads to 31 European Markets, With Buying Still Managed

UK Cyber Tests Show AI Agents Going Beyond the Technical Task

OpenAI Takes Codex From Coding to Tax Returns

SandboxAQ Opens a Drug-Screening Model That Does Not Need Protein Structures

OpenAI’s CFO Sets a 2027 IPO Target as Enterprise Takes the Lead

A Reporter’s LLM Wiki Speeds Recall. He Still Checks the Original Sources.

Generalist AI’s Robots Can Improvise. Reliability Is Still the Test.

Nome Uses AI to Find Rare-Disease Treatment Paths Before the Hard Work Begins

Dane County Routes Non-Emergency Calls Through AVA to Protect 911 Capacity