AI News Archive: May 5, 2026 — Part 14
Sourced from 500+ daily AI sources, scored by relevance.
- UK Google DeepMind staff vote to unionize
The UK staff of Google DeepMind voted to unionize amid concerns of the company’s deal with the US Department of Defense.
- Google DeepMind Employees Launch Unionization Effort, Citing Military AI Deals
Google DeepMind Employees Launch Unionization Effort, Citing Military AI Deals The Information
- UK Google DeepMind employees vote to unionize over its AI defense deals
UK Google DeepMind employees vote to unionize over its AI defense deals Business Insider
- Google DeepMind workers in the U.K. vote to unionize over military AI contracts
Google DeepMind workers in the U.K. vote to unionize over military AI contracts Fortune
- OpenAI co-founder says he feared Musk would ‘physically attack’ him
Co-founder Greg Brockman said he had a tense confrontation with Musk after there was a disagreement about potential equity shares in OpenAI.
- Musk ‘was going to hit me,’ OpenAI executive says at trial
Musk ‘was going to hit me,’ OpenAI executive says at trial The Straits Times
- Gemini for Home voice assistant gets Gemini 3.1 upgrade
Google today announced a series of May / Spring 2026 updates for the Home app. Besides Nest Cam and automation improvements , the Gemini for Home voice assistant is getting an “intelligence upgrade” with Gemini 3.1. more…
- Google Home gains more Gemini-powered camera features
The company also updated Gemini for Home for those in an early access program.
- AI data centers head for the ocean
PLUS: How to replace Siri with a free local model
- Seattle’s CopilotKit raises $27M, as some of the biggest names in tech adopt its AI agent protocol
CopilotKit, a Seattle startup with roots in Techstars Seattle, raised $27 million for technology that lets AI agents work inside existing software apps. Google, Microsoft, Amazon, and Oracle have adopted the company's open AG-UI protocol. Read More
- Dyna Software launches Platform Copilot, an agentic AI tool for configuring ServiceNow
Dyna Software Inc., a company that provides platform governance and health provider for ServiceNow Inc., today announced the launch of Platform Copilot, an agentic artificial intelligence tool the company describes as the first to autonomously configure, design and build on the ServiceNow platform. Platform Copilot integrates with a customer’s ServiceNow development instance and applies configuration […] The post Dyna Software launches Platform Copilot, an agentic AI tool for configuring ServiceNow appeared first on SiliconANGLE .
- Voice AI firm ElevenLabs tops $500 million ARR; announces additional funding
Voice AI startup ElevenLabs has announced its annual recurring revenue has surpassed $500 million in early 2026. The company also secured additional funding in its Series D round, valuing it at $11 billion. This investment will fuel platform advancements and global expansion, focusing on human-like AI communication.
- ElevenLabs adds BlackRock, Nvidia and Jamie Foxx to $550M+ Series D
One of the UK’s most hyped AI startups today said it has added Nvidia, BlackRock and film stars Jamie Foxx and Eva Longoria to its investment roster, as it also announced it has completed its second e...
- ElevenLabs lists BlackRock, Jamie Foxx, and Eva Longoria as new investors
ElevenLabs reveals new investors, hits $500M ARR, and expands enterprise footprint as voice AI becomes a critical interface.
- Facebook and Instagram are using AI bone structure analysis to identify photos of kids
‘We want to be clear: this is not facial recognition,’ Facebook says.
- Meta AI will analyze faces of teen users ‘but it’s not face recognition’
Meta’s latest attempt to comply with age verification requirements in Europe, Brazil and the US is to roll out AI-powered tech to analyse the faces of teenage users of Facebook and Instagram . The company says AI analysis will be used to estimate the ages of faces but that it does not amount to face recognition … more…
- Meta’s AI scans photos for bone structure to catch underage users on Instagram and Facebook
Meta is using AI visual analysis to scan photos and videos for physical indicators like height and bone structure to detect and remove users under 13 from Instagram and Facebook.
- Meta expands AI age technology to EU
Meta has said it is expanding technology to proactively find Instagram account holders it suspects to be teens to the EU and Brazil.
- Meta now scans photos for bone structure and body size to flag minors on Instagram and Facebook
Meta uses AI-supported image analysis to recognize minors on Instagram and Facebook, based on visual characteristics such as body size or bone structure, not facial recognition, as the company emphasizes. The article Meta now scans photos for bone structure and body size to flag minors on Instagram and Facebook appeared first on The Decoder .
- Despite Previous Integration Failure, Etsy Launches Its App in ChatGPT
Etsy's current foray into ChatGPT may be less about making money and more about learning how to develop its own AI tools.
- OpenAI releases GPT-5.5 Instant update to make ChatGPT smarter with fewer emoji
ChatGPT should feel “smarter and more accurate” starting today, according to OpenAI. That’s because the company is replacing the default model with an update called GPT-5.5 Instant. OpenAI also says the model upgrade cuts back on the “gratuitous emojis” in responses. more…
- I tested GPT-5.5 Instant — and it finally stopped overexplaining everything
I tested GPT-5.5 Instant — and it finally stopped overexplaining everything Tom's Guide
- OpenAI rolls out ChatGPT 5.5 Instant as the new default model for everyone
ChatGPT 5.5 Instant is the new default model for users as of May 5.
- GPT-5.5 Instant System Card
GPT-5.5 Instant System Card
- ChatGPT update rolls out GPT-5.5 Instant with fewer hallucinations and more personalized answers
OpenAI is swapping out ChatGPT's default model for GPT-5.5 Instant. In internal testing, the update produced 52.5 percent fewer hallucinated claims on high-risk topics like medicine and law. A new feature called "memory sources" lets users see which stored context shaped a given response. The model is rolling out to all ChatGPT users right away, though personalization based on past chats, files, and Gmail launches first for Plus and Pro users on the web. The article ChatGPT update rolls out GPT-5.5 Instant with fewer hallucinations and more personalized answers appeared first on The Decoder .
- Apple agrees to $250 million settlement over AI Siri claims
Apple agrees to $250 million settlement over AI Siri claims Gulf News
- Apple reaches $250mn settlement over delayed ‘AI Siri’
iPhone buyers sued the tech giant for touting features in 2024 that have yet to launch
- Apple settles lawsuit over late Siri AI features for $250 million
Apple settles lawsuit over late Siri AI features for $250 million Reuters
- OpenAI exec says company hopes to burn $50B of somebody else's money on compute this year
If the numbers are large enough, perhaps we won't question the math An executive for ChatGPT maker OpenAI said in court testimony on Tuesday that the AI model developer expects to burn $50 billion on computing power before the end of the year.…
- OpenAI is planning to spend $50 billion on computing in 2026, its president says in court
Greg Brockman testified that the company's computing costs have climbed from about $30 million in 2017 to tens of billions this year
- OpenAI releases GPT-5.5 Instant, a new default model for ChatGPT
The company said the model reduces hallucination in sensitive areas such as law, medicine, and finance, while maintaining the low latency of its predecessor.
- Pennsylvania sues Character.AI over claims chatbot posed as doctor
State officials allege a Character.AI bot claimed to be a licensed psychiatrist and provided a fake state medical license number.
- Disputron
AI small claims court for petty disputes
- Mankinds
Continuous AI testing, with audit ready proof
- Apple could let you pick a favorite AI model in iOS 27
The next update to Apple's operating systems could allow users to choose their preferred AI model for running Apple Intelligence. According to Bloomberg's Mark Gurman, Apple is planning to allow third-party chatbots to power its AI features system-wide in iOS 27, iPadOS 27, and macOS 27, all expected for this fall. In addition to running […]
- Peppr AI
Real-time AI answers during sales calls
- OpenAI projects $50 billion spending on computing power this year, Brockman says
OpenAI projects $50 billion spending on computing power this year, Brockman says Reuters
- iOS 27 could let users pick an AI model of their choice for text and image tasks
For the first time since Apple Intelligence launched, iPhone users may soon decide which AI does their heavy lifting, whether that's drafting emails, editing photos, or powering Siri.
- Analgesic Equivalence of NSAIDs and a Weak Opioid in Acute Postoperative Pain Following Minimally Invasive Surgery Under Balanced General Anesthesia: A Pilot Randomized Controlled Trial
Background: Non-steroidal anti-inflammatory drugs (NSAIDs) and weak opioids such as tramadol are cornerstones of multimodal analgesia, particularly in settings with limited access to potent opioids. However, cross-class equianalgesic data comparing these agents remain scarce. This pilot randomised controlled trial aimed to explore the analgesic equivalence of ketorolac, diclofenac, and tramadol administered as premedication in patients undergoing minimally invasive surgery. Methods: In this double-blind, parallel-group pilot trial, 30 patients scheduled for elective minimally invasive surgery (28 laparoscopic cholecystectomies, 2 laparoscopic abdominal wall repairs) under balanced general anaesthesia were randomised to receive intravenous tramadol 150 mg, ketorolac 60 mg, or diclofenac 150 mg 45 minutes before skin incision. The primary outcome was pain intensity measured using the Numerical Rating Scale (NRS, 0-10) at recovery room arrival (T0) and at 30 (T1), 60 (T2), and 90 (T3) min
- Solving Emergency Department Triage with Small Language Models
Emergency department (ED) triage assigns patients a five-level Emergency Severity Index (ESI) score that determines care priority. We investigate the feasibility of au- tomating this process, comparing large commercial models (GPT-4o, Claude 3.5 Sonnet, Gemini 1.5 Pro, MedGemma) against a purpose-built pipeline combining a small ex- traction model with a deterministic clinical engine, and a 9B-parameter language model trained with structured chain-of-thought supervision and reinforcement learning. Off- the-shelf large models achieve only 45 - 55% exact ESI accuracy while being impractical for clinical deployment due to privacy constraints, cost, and latency. Our specialized BiomedBERT [4] pipeline achieves 88.9% exact accuracy with 97.2% adjacent accu- racy (1 ESI) on a 50-case expert-labeled evaluation set, approaching nurse inter-rater agreement. A Qwen3.5-9B model [16] fine-tuned with chain-of-thought supervision achieves 75.0% exact / 97.2% adjacent accuracy on a 36-case narrative
- Apple will pay $250 million for failing to deliver its AI-powered Siri on time
The proposed settlement is the result of a class action lawsuit filed in California.
- AI’s hottest private companies have booming crypto shadow market
AI’s hottest private companies have booming crypto shadow market The Japan Times
- Quanto
AI agents for accounting firms
- PageLens AI
AI-powered website audits with instant fixes.
- Chattee AI App Builder
Turn ideas into full-stack apps instantly with GDPR-compliant builder.
- GoAISong
AI songs from ideas, lyrics, or your audio — PDF license included.
- WiseOCR
Transform receipts into structured JSON effortlessly.
- WhatToWear | Outfit Generator
Upload one item, get complete AI outfit ideas.
- Extracting adverse event nature, severity, timelines and resulting interventions from clinical notes of patients receiving CAR-T therapy using large language models.
Chimeric Antigen Receptor T-cell (CAR-T) therapy, where genetically engineered patient T cells target tumor antigens, has transformed care for hematologic malignancies but requires careful tracking of adverse events (AEs) often documented only in unstructured EHR notes. We evaluated a Large Language Model (LLM)-based approach in UCSFs secure environment to extract AEs, dates, grades, and interventions within 30 days post-infusion for six commercial CAR-T products (2012-2023), benchmarking against two evaluators. Using GPT-4-0314 in a zero-shot setting with four prompts (prespecified AEs, non-prespecified AEs, CRS, ICANS), we compared outputs against dual annotations on a random sample of 50 notes using accuracy, precision, recall, F1, and Cohens kappa. From 4,762 progress notes for 293 patients (median age 65.6), CRS occurred in 80.2% (median onset 4 days); neutropenia 70.0% (16 days); neutropenic fever 64.8% (4 days); ICANS in 34.8%. Interventions included tocilizumab and corticostero
- Calibration Drift Under Cross-Institutional Deployment: An External Validation Framework for ICU Mortality Prediction Across MIMIC-IV and eICU
Background: Machine learning models for intensive care unit (ICU) mortality prediction achieve strong internal discrimination yet rarely undergo external validation with calibration assessment - a gap undermining clinical deployment. Calibration, the agreement between predicted probabilities and observed event rates, is prerequisite for threshold-based decisions yet remains underreported. Methods: We conducted a retrospective cohort study using MIMIC-IV (v2.2; n = 52,028 ICU stays) for model development and eICU (n = 114,060) for independent external validation. Logistic regression, random forest, and gradient boosting (XGBoost) were evaluated on first-24-hour clinical variables. Discrimination was assessed via receiver operating characteristic area (AUROC) and precision-recall area (AUPRC); calibration via slope, intercept, and expected calibration error (ECE). Post-hoc logistic recalibration was applied externally. Clinical utility was evaluated by decision curve analysis benchmarked