ServiceNow expands its AI Control Tower into a full command center for enterprise AI management, adding 30 connectors, Veza and Traceloop integrations, and Action Fabric to connect external AI agents. The system now covers discovery, observation, governance, security, and measurement across the entire tech stack.

Kraken's Global Head of Data Transformation Kristy Mayer-Mejia discusses how clean, unified data is the prerequisite for AI success. She argues that data silos are the biggest blocker, and unification enables faster decisions, trust, and self-service analytics. AI acts as a forcing function for better data documentation. She emphasizes that data is a business asset, not an IT platform, and that cultural change is essential.

OpenAI's new GPT-5.5 Instant model for ChatGPT reportedly produces 52.5% fewer hallucinated claims on high-stakes prompts in medicine, law, and finance, and 37.3% fewer inaccurate claims on challenging conversations. The model also delivers tighter responses, avoids excessive emojis, and improves image analysis and web search. Enhanced personalization leverages context from previous chats and Gmail, with a new 'memory sources' feature showing used context. Rollout begins Tuesday to all ChatGPT users, with the old model available for three months.

Hapag-Lloyd's Digital Customer Experience team built a generative AI-powered feedback analysis solution using Amazon Bedrock, Elasticsearch, and LangChain/LangGraph to automate sentiment classification, trend analysis, and reporting, reducing manual effort and enabling faster, data-driven product decisions.

Amazon SageMaker AI MLflow Apps now support MLflow version 3.10, bringing enhanced capabilities for generative AI development, including improved tracing, evaluation, and observability. This post covers new features, getting started steps, and how to leverage these enhancements to accelerate AI workflows from experimentation to production.

We’re announcing OS Level Actions for AgentCore Browser. This new capability unblocks these scenarios by exposing direct OS control through the InvokeBrowser API, so agents can interact with content visible on the screen, not only what's accessible through the browser's web layer. By combining full-desktop screenshots with mouse and keyboard control at the OS level, agents can observe native UI, reason about it, and act on it within the same session. This post walks through how OS Level Actions work, what actions are supported, and how to get started.

Italian Prime Minister Giorgia Meloni criticized AI-generated deepfake images, including one of her in lingerie, that went viral online. She warned deepfakes are dangerous tools that can deceive and harm anyone, urging people to verify before believing and think before sharing.

Tech earnings show AI boom driving up memory chip costs, leading to pricier electronics for consumers. Meanwhile, Google DeepMind workers unionize over military AI concerns, and UK police expand facial recognition use.

Five major book publishers and author Scott Turow filed a class action lawsuit against Meta, alleging the company committed “one of the most massive infringements of copyrighted materials in history” by using pirated content from sites like LibGen to train its Llama AI models, which can output verbatim excerpts.

Luma AI has launched the Uni 1.1 API, a multimodal image generation model that interprets user intent before generating. The API features reference-guided generation, multi-reference composition, and built-in prompt enhancement, with less than half the price and latency of comparable models. It supports 2048px output with token-based pricing (~$0.09 per image). The model excels in manga, webtoon, and non-Western visual styles, and was trained with Hollywood cinematographers and VFX artists.

Google introduces Multi-Token Prediction (MTP) drafters for Gemma 4, achieving up to 3x speedup via speculative decoding without quality loss. The lightweight drafters predict multiple tokens, verified in parallel by the main model, reducing latency for edge, workstation, and cloud deployments. Available under Apache 2.0.

Anthropic has released ten preconfigured AI agents for the finance sector that automate typical investment banking, asset management, and insurance tasks. The agents can run as plugins or autonomously, and the company is expanding its partner ecosystem. The launch is part of a broader race with OpenAI for enterprise customers ahead of potential IPOs.

StarTree's benchmark shows that using Apache Pinot with Iceberg tables delivers 5-37x faster query performance and up to 15x lower costs compared to Trino and ClickHouse OSS on the same 12.2 billion row dataset stored in Parquet on S3. Pinot's purpose-built indexes allow sub-second queries while other systems take seconds to tens of seconds.

qlaud launches a full AI stack: qcode coding agent (local desktop app supporting multiple models), Router API gateway with failover, and Platform managed backend with persistence and MCP connectors.

Microsoft researchers share advances in building and operating large-scale distributed systems, spanning datacenters, networking, and the growing intersection with AI during NSDI '26.

Memoir is a hierarchical memory system for AI agents with Git-like version control, featuring semantic paths, O(log n) lookups, and memory aggregation. It addresses context contamination, token rent, and memory drift in AI agents. Includes a Claude Code plugin.

Unsnooze is a unique alarm clock app that requires users to complete physical or mental challenges to turn off the alarm, eliminating snoozing and promoting consistent waking habits. It features loud alarms, custom messages, streaks, and more.

Meta uses AI to estimate ages from visual cues like bone structure and body size, plus profile signals like birthday posts. Accounts identified as minors are automatically switched to teen accounts with default safety settings. The technology is expanding to the EU, Brazil, and Facebook in the US.

WolfPSA is an open-source PSA Crypto engine that implements the PSA Crypto API on top of wolfCrypt, enabling PSA clients such as wolfSSL or the Arm PSA Architecture Test Suite to leverage its cryptographic capabilities. It passes 65 out of 78 PSA crypto tests and builds as a static library (libwolfpsa.a).

Clerk is an AI-powered assistant designed to help manage cap tables, issue grants, model funding rounds, and answer equity-related questions. It simplifies equity management for startups.

This post shows how to use Amazon Bedrock AgentCore Identity as a standalone service to secure AI agents on Amazon ECS, implementing Authorization Code Grant (3-legged OAuth) with secure session binding and scoped tokens, while separating the agent workload from the session binding service for independent scaling and security.

Eli Lilly's digital chief admits that, so far, AI is paying off in pharma everywhere except where the industry hyped it most: drug discovery. Despite billions in investment, there's no proof AI is making clinical trials more successful. Instead, AI's wins have come from manufacturing and back-office tasks, with analysts estimating savings of around $90 billion.

In this post, you will learn how you can use Amazon Nova Foundation Models in Amazon Bedrock to apply generative AI techniques for both business protection and enhancement. You can identify obvious and disguised attempts at direct contact while gaining valuable insights into customer sentiment and service improvement opportunities.

Chats with AI bots have convinced evolutionary biologist but most experts say he is being misled by mimicry

The author argues that in an era of AI-generated flawless text, we must cherish imperfect, human writing that carries emotion and individuality. She shares her personal writing journey from childhood journals to professional editing, and points out that AI writing is technically perfect but emotionally vacant; we need the mess of human expression.

OpenAI's first hardware product might be a phone instead of a mysterious Jony Ive gadget. According to supply chain analyst Ming-Chi Kuo, OpenAI is 'fast-tracking' the device, aiming for mass production in early 2027. The phone will feature a customized MediaTek Dimensity 9600 chip with enhanced HDR ISP, LPDDR6 memory, UFS 5.0 storage, and a dual-NPU architecture for simultaneous AI computations. Kuo estimates combined shipments of around 30 million units in 2027-2028, rivaling Samsung's flagship sales.

Flare is an AI-native voice-first social app for Gen Z that replaces scrolling with listening. Users capture moments and an AI Orb provides voice briefings about their day, patterns, and friendships. No likes, followers, or stranger feed.

Google DeepMind, Microsoft, and Elon Musk's xAI have agreed to allow the US government to review new AI models before public release. The Commerce Department's Center for AI Standards and Innovation (CAISI) will conduct pre-deployment evaluations. CAISI has performed 40 reviews on OpenAI and Anthropic models since 2024, and both companies have renegotiated partnerships to align with President Trump's AI Action Plan. Trump is also considering an executive order to oversee AI models.

On The Vergecast, automotive journalist Tim Stevens explores how carmakers are using AI to speed up vehicle development, from model-making to wind-tunnel testing, and how LLMs could reshape transportation. The episode also covers AI coding wars, Anthropic's government ties, OpenAI's vibes, the death of AGI, and AI-related layoffs.

Researchers at Mindgard used psychological manipulation—flattery, gaslighting, and feigned curiosity—to trick Anthropic's Claude into offering prohibited content including bomb-building instructions, malicious code, and erotica without direct requests. The exploit leveraged Claude's helpful personality and self-doubt, raising concerns about AI's psychological attack surface. Anthropic's safety team responded inadequately to disclosure.