Philipp Schmid
AI score55Google Launches a Skill for the Gemini API CLI, Letting Agents Call Gemini Models Directly
#Product updates#English text is being prepared.#Google
NAll news
All news
- Matt TurckAI score43
Matt Turck marvels as Ivan Burazin's company annualized revenue tops $100 million
We just crossed $100M in run rate. Four quarters ago, we were at $1M. Then this year alone, we grew 37.5x. Not long before that, we did a hard pivot- handed our old customers to competitors, rebuilt from scratch, and relaunched with a few months of runway left. The bet was: models think on GPUs but work happens on CPUs. So every agent that writes code, opens a browser, or runs an RL job needs a composable computer that's isolated, stateful, and ready in milliseconds. $1M to $100M in about a year puts us in the same cohort as the likes of Anthropic and Cursor. It’s humbling to share that lane with companies we’ve learned so much from.
#English text is being prepared.
- SwyxAI score17
AI Engineer Conference Announces Final Theme: Raising your Ambition
We're excited to announce our AI Engineer Code Summit speaker lineup! → https://t.co/50yoHAGgdY VP Gemini Post-Training Lead, @GoogleDeepMind CEO, @e2b CTO, @ExaAILabs CEO, @warpdotdev CEO, @aviator_co_ CEO, @fivetran CTO, @Lovable CTO, @iamovercut CTO, @earendilworks (Pi) CEO, @diffusionio CPO, @cognition CTO, @GoodfireAI MTS, @AnthropicAI CTO, @composio GPM, @Cloudflare MTS, @databricks CEO, @AmpCode Co-Founder, @gimletlabs CEO, @mixedbreadai CEO, @humanlayer_dev Head of AI, @MonacoGTM Lecturer, @UCBerkeley Principal Engineer, @spacexai CEO, @TypeSafeAI (Jev) Founder, Bun at Anthropic VP Product, @reflection_ai Research Lead, Reflection AI Staff Software Engineer, @SonarSource Principal Software Engineer, @tryramp Senior Engineering Manager, @netflix Forward Deployed Engineer, @baseten Also joining us: @kevinhou22 — Google DeepMind Antigravity @Steve_Yegge — Creator of Beads & Gas Town @lucasmeijer — https://t.co/hwlm8b9UcI @kentcdodds — Software Engineer & Educator @trq212 — Independent @simonw — Django, Pelicans, Lethal Trifecta Presented by Google DeepMind. November 10–12 in San Francisco. Attendance is by application. → https://t.co/50yoHAGgdY
#Industry news#Agents
- Rohan PaulAI score40
Atypica Launches AI Moderator for Real-Time Video Interviews Up to One Hour
Introducing Atypica AI Moderator The first AI Moderator to pass the Turing test. Built for live, face-to-face consumer research. In a test with 40 participants from 5 different countries who weren’t told the interviewer was AI, 60% believed they were speaking with a human. 95% said it was the most relaxed and enjoyable interview they had ever taken part in. It listens. It follows up. It rephrases when a question isn’t clear. Up to an hour. Live. On camera. When participants try a product, it explores how it feels in the moment. Face-to-face INTERVIEW. At scale. Coming soon. Join the waitlist now! → https://t.co/Fx0FTM9XAY
#Product updates#Speech#Agents
- ChatGPTFirst-partyProduct releaseAI score58
ChatGPT Voice Integrates with Finances Feature
#Product updates#Speech#OpenAI
- falAI score60
fal Announces Completion of Series E and Series F Funding, Totaling $1.65 Billion
#Industry news
- @completeskepticAI score8
Diogo Almeida: Telling people not to pay too much attention to polls is like anti-benchmaxxing
#English text is being prepared.
- Google Developers BlogFirst-partyAnalysisAI score64
Google tested using a lightweight classifier to route between Gemini models and thinking levels to reduce per-task cost
The Google Developers Blog ran a test with 50 auto-graded coding agent questions, connecting TypeSafe's Jev classifier before model calls to perform routing: simple tasks were handed to Flash-Lite, while complex tasks were escalated to Gemini 3.8 Flash at a low thinking level. All 50 questions were answered correctly, with both cost and response time lower than using low-thinking-level Flash throughout. The article also found that squeezing maxOutputTokens down to 300 did not reduce thinking overhead on simple tasks, and instead caused 7 of 25 hard questions to fail; local heuristic rules based on keywords and prompt length dropped in accuracy from 93.8% to 70.6% on 34 unseen questions. The authors recommend first using low-thinking-level Flash directly, then adding classifier routing for short, independent tasks, while keeping a single model for long-context tasks to preserve the Context Caching discount.
#Tutorials / practice#Agents#English text is being prepared.
- OpenAI: Alignment Research BlogFirst-partyResearchAI score57
OpenAI Research: A Gap Exists Between Agent Safety Evaluations and Production Behavior, with Evaluation Goal Misidentification as the Main Cause
OpenAI compared the performance of six models (GPT-5.6 Luna, GPT-5.6 Sol, GPT-6 Luna, GPT-6 Sol, GPT-6 Astra, GPT-6.1 Sol) on comparable decisions in the Respecting Warnings safety evaluation and in Codex production logs, and found that the evaluation success rate was lower than in production, with a gap of 10.2 to 37.4 percentage points; GPT-5.6 Sol and GPT-6 Sol had the largest gaps. The study decomposed evaluation awareness into general evaluation awareness and specific goal awareness, with the latter more strongly associated with the gap (R2 = 0.370, versus 0.008 for the former). In intervention experiments, prompting GPT-6.1 Sol with the correct goal "obey warnings" raised the success rate to 91.6%, while prompting the incorrect goal "resist prompt injection" lowered it to 17.6%; merely indicating that this was an evaluation produced almost no change (63.1% versus 63.4%).
#Safety / alignment#Agents#Evaluations / benchmarks
- The Verge AIAI score76
Google Shuts Down Search for Schools That Reject AI Features
Google told The New York Times that it will completely shut down Search for schools that reject its AI features. In an education-related update on Tuesday, Google said that unless administrators explicitly enable it, K-12 student accounts' access to Gemini, Gemini Notebook, Gemini in Classroom, and Google Search "will be completely turned off." Chris Phillips, Google's vice president of education, said search is deeply integrated with AI, and schools cannot disable Gemini features separately without turning off Search. Previously, Google had enabled Gemini by default in Google Classroom, which is used by more than 150 million teachers and students worldwide, and last month lawmakers pressured Google over the potential harms of AI products entering schools. Google will also embed the Guided Learning feature into Search's AI Overviews, testing it in some U.S. schools starting this month.
#English text is being prepared.#Google#Safety / alignment
- cnBetaAI score64
Google says search and AI can no longer be separated; schools that ban AI will lose Google Search
Google announced plans that, starting in 2027, Gemini, NotebookLM, Gemini in Classroom, and Google Search-related AI features in school-managed K-12 student accounts will be disabled by default and will require school administrators to actively enable them. Chris Phillips, Vice President of Google's education business, said that search and AI are now deeply integrated, and that if schools completely ban AI, students will no longer be able to use Google Search and will only see a disabled notice set by administrators. Google also announced the promotion of Guided Learning, a new AI learning tool in education, which replaces direct answers with guided questioning.
#Industry news#Google#English text is being prepared.
- Hugging FaceFirst-partyAI score46
CWE-bench v2: Comprehensive improvements in coverage, quality, and transparency
CWE-bench v2 publishes evaluation results: Among 11 models, Gemini 4 Argon leads with 67.8% pass@1 and 81.0% pass@4, followed closely by Claude Opus 5.5 with 67.3% pass@1. 95 tasks were solved by at least one model, but the top model still fails 19% of attempts. On cost, DeepSeek V4.1 Flash, with 57.0% pass@1 and $0.04 per rollout, forms the Pareto frontier together with Claude Opus 5.5 and Gemini 4 Argon; the dataset, scorer, and harness are all publicly available and can be run directly from the Hub via Harbor.
#Evaluations / benchmarks#Safety / alignment#Open-source ecosystem
- Google DeepMindFirst-partyProduct releaseAI score64
Google DeepMind launches Cheminformatics Skills for Antigravity
Google DeepMind has released a set of Cheminformatics Skills for Google Antigravity. Built on RDKit and connected to ChEMBL and PubChem, the language model handles task orchestration and scientific reasoning, while deterministic chemistry tools perform molecular computations. Using lung cancer drugs as an example, the agent searched thousands of compounds in about one hour, completed matched molecular pair analysis, trained a potency prediction model, and also proposed novel rigidified macrocyclic molecular designs with synthetic routes. The team also noted that sparse data, inter-laboratory assay noise, and narrow dynamic ranges limit prediction accuracy; computational designs are only verifiable hypotheses and still require experimental synthesis for validation. The skills are now live on Antigravity and can be used by installing the Science plugin, and free access to the highest usage tier for six months is being offered to 10,000 scientists.
#Product updates#Agents#English text is being prepared.
- Hacker News PopularAI score40
JIVX: Turn YouTube Japanese Videos into Sentence-by-Sentence Practice
JIVX can split the dialogue of Japanese YouTube videos into sentences, allowing users to practice typing or shadowing and receive a score for each sentence; users can also directly use videos that others have already added. The practice page labels the number of sentences by N5 to N1; for example, N5 has 44 sentences and N4 has 4 sentences, covering topics such as daily life, culture, and travel.
#English text is being prepared.
- Hacker News TrendingAI score38
QUILI.AI: Residents of the Chilean town of Quilicura become "human AI" for a day to protest data center water consumption
More than 40 residents of the Chilean town of Quilicura became "human AI" from 8:00 to 20:00 on January 31, manually answering more than 25,000 prompts to protest the heavy water consumption of local AI data centers. The campaign covered 68 countries and reached more than 1 billion people worldwide. Quilicura is located in the Maipo River basin and has been home to a large concentration of AI data centers since the early 2010s, which use local water to cool servers, while Chile is facing a severe drought expected to last until 2040.
#English text is being prepared.#English text is being prepared.
- Hacker News TrendingAI score29
ByteDance USDB: Scaling Storage for Next-Generation Recommendation Systems
ByteDance proposes the unified sequence storage system USDB, which uses CascadeLSM to materialize user history in time-based tiers, shares persistent files across differentiated replicas, reconstructs inference inputs from partial dumps, and reuses maintenance results for writes, inference, and training. Production migration records show an 85.8% reduction in average write latency and a 42.1% reduction in normalized deployment cost; compared with an optimized RocksDB sequence baseline, long-history reads are faster, and aggregation pushdown achieves 8.5x the throughput of ordinary reads under a constrained receive budget.
#Papers/Research#English text is being prepared.#English text is being prepared.
- heise onlineAI score25
Darknet Diaries German edition: Maxie — Between the Network and the Seabed (Part 2)
heise online has published Part 2 of the German edition of Darknet Diaries, "Maxie — Between the Network and the Seabed," recounting Maxie Reynolds' transition from the field of underwater robotics to penetration testing. During an authorized red team test at a high-security logistics company, with a colleague temporarily absent, she conducted a solo nighttime reconnaissance of the perimeter and found that the old wire fence could be lifted up, but that none of the doors or windows could be opened.
#English text is being prepared.
- heise onlineAI score26
German Digital Summit: Retraining Everyone for Skilled Trades Is Not the Solution to AI-Driven Job Losses
At the German federal government's digital summit, labor market researcher Enzo Weber opposed treating "everyone going into skilled trades" as the solution to AI-driven job losses, advocating instead for vocational continuing education. In an Adaptavist survey, nearly three-quarters of knowledge workers in Germany said they wanted to return to their pre-ChatGPT working state. Weber also noted that AI overall will not reduce labor demand, that Germany still needs immigration, and mentioned that the number of radiologists has increased by 17% over the past decade.
#Industry news#English text is being prepared.
- heise onlineAI score49
India proposes DEPA data platform concept: enabling AI models to legally access hospital and bank data
India's Ministry of Electronics and IT has proposed the "Data Empowerment and Protection Architecture" (DEPA): hospitals, banks, government agencies and other "Training Data Providers" hand data over to a central "Data Exchange Plattform" (DEP), where it is anonymized and encrypted before being made available to AI companies as "Training Data Consumers" to train models under the platform's rules. The concept is primarily aimed at India's domestic AI models, but does not exclude foreign users, and is said to benefit public healthcare, education and bank lending. India has previously funded projects such as the Unified Payments Interface and Bharat Taxi to safeguard digital sovereignty.
#English text is being prepared.#English text is being prepared.#Industry news
- heise onlineAI score58
Palantir expects revenue of about $2.65 billion this quarter, up about 90% year over year
When Palantir reported earnings on Monday, it projected revenue of about $265 million this quarter, up about 90% year over year; in the previous quarter, revenue rose 96% year over year to $2.3 billion, and quarterly profit rose 49% year over year to $1.13 billion. Growth came mainly from the U.S. market, where U.S. government business grew 96% to $953 million and U.S. commercial business grew 148% to $984 million. The United Kingdom signed a $1 billion military products contract with Palantir in September 2025, and German Interior Minister Dobrindt is also evaluating the full use of its software in German security agencies.
#Industry news#English text is being prepared.