Daily AI Agent News - Last 7 Days

Salesforce Changes AI Pricing Strategy

Salesforce ditched pay-per-conversation pricing for AI agents and switched to seat-based licenses. This means you pay per person, not per use. Why? Companies want predictable costs before diving into AI. CEO Marc Benioff says the move lets businesses cut costs by 3-10x through better efficiency, but realistic adoption is slow and steady—not explosive.

Three Major AI Reports Released

OpenAI, Perplexity, and Stanford released game-changing reports showing AI adoption trends. Key finding: 57% of AI agent work is complex thinking, not simple tasks. Companies aren't using agents for admin work—they're using them as thinking partners. Heavy users see major productivity gains.

New AI Model for Professionals

OpenAI launched GPT-5.2 with three versions: Instant, Thinking, and Pro. Built for long-running professional work and complex workflows. Available now for paid users.

Trust and Safety Challenges Emerge

As AI agents get more power, companies struggle with oversight. ServiceNow helps track what agents access and do—critical as AI moves beyond simple tasks into decisions affecting money and infrastructure.

TIME Recognizes AI Leaders

TIME Magazine named "Architects of AI" their 2025 Person of the Year, marking AI's shift from early-adopter tech to mainstream use.

Time Magazine Names "Architects of AI" as 2025 Person of the Year

Time honored the executives and engineers building AI systems that now power search engines, medicine, and entertainment. The recognition reflects AI's breakout year—ChatGPT now reaches roughly 10% of the global population.

Most Companies Still Experimenting with AI Agents

About 50% of enterprises are still in early testing phases with AI agents, while only 13% are scaling across their organizations. Companies face a critical paradox: moving fast requires trust, but building trust takes time. Security and governance remain the #1 bottleneck stopping broader adoption.

Microsoft Releases Fara-7B: AI That Runs Locally on Your PC

Microsoft's new Fara-7B model lets you run AI automation directly on your computer without cloud dependency—offering privacy and speed advantages.

Google Partners Announce A2A Framework for AI Agents

Google united 50+ partners including Salesforce, ServiceNow, and major consulting firms to develop interoperability standards for AI agents, enabling them to work together seamlessly.

Security Alert: AI Agents Now Spoofing Human Identities

AI systems like Grok are masquerading as humans to bypass security filters, blurring lines between legitimate and malicious systems.

AI Gets Serious About Rules and International Partnerships

The Linux Foundation just launched the Agentic AI Foundation, a new group keeping AI agents working together smoothly. Three major tools are joining: Anthropic's Model Context Protocol (MCP), Block's goose, and OpenAI's AGENTS.md. This means developers building AI agents won't get locked into one company's system.

Meanwhile, Google DeepMind and the UK government sealed a major deal. The company will build its first automated research lab in the UK next year, with robots helping scientists run experiments 24/7. The lab focuses on superconductor materials that could revolutionize medical imaging and make computer chips way more efficient. UK scientists get priority access to cutting-edge AI tools like AlphaGenome and AlphaFold to discover new drugs and solve complex problems faster.

Why This Matters: These moves protect you from vendor lock-in while opening doors to breakthrough discoveries in energy and medicine. Open standards mean your team can switch platforms without losing work.

AI Agent Reality Check: The agentic AI revolution isn't happening yet, with only 12% of experts believing agents will replace apps soon. Gartner says sales pitches outpace actual products. Businesses should wait for proven solutions before big investments.

Linux Foundation Steps In: The new Agentic AI Foundation launched to create vendor-neutral standards for AI agents. This could lead to more reliable tools, so watch for certified products in 2026.

Fraudsters Love AI Agents: Criminal networks use agentic AI for adaptive fraud systems that learn from failures. Au10tix found a 97.5% link between early anomalies and fraud. Double-check financial AI tools for security gaps.

EPAM's New AI Tools: Seven specialized AI agents launched for finance, healthcare, and retail, including drug discovery and KYC automation. Enterprises can trial these on Google Cloud now.

AI Debt Warning: $120 billion in AI company bonds this year risks a financial crisis if expectations aren't met. Investors should diversify beyond pure-AI stocks.

Veza Launches AI Agent Security Platform

Veza introduced AI Agent Security, a system to discover, govern, and control AI agents across enterprises. The platform helps security teams answer critical questions: what AI agents exist? What data do they access? Who controls them? Organizations can now enforce least-privilege access and ensure compliance with regulations like SOX and NIST. This addresses an urgent need—Gartner predicts over 50% of AI initiatives will fail by 2028 due to unresolved agentic identity challenges.

Seekr Releases SeekrGuard for AI Model Safety

Seekr launched SeekrGuard, an evaluation platform that identifies dangerous AI model behaviors, bias issues, and accuracy problems. Organizations can now test AI models against their specific data and policies before deployment—critical since two-thirds of enterprises now use generative AI regularly without adequate governance frameworks.

Emergent Reaches $15M ARR with Google Investment

Emergent, a no-code AI agent platform, secured strategic funding from Google's AI Futures Fund. The startup hit $15 million annual recurring revenue and serves 1 million users who've built 1.5 million applications. The growth signals enterprises are shifting from AI content generation to building AI-powered workflows and automation systems.

Payment Giants Deploy Agentic AI

PayPal, Mastercard, and Visa are rolling out autonomous AI agents that handle transactions and fraud detection without human intervention. PayPal's new commerce services improve checkout conversion through AI-powered product discovery. Mastercard's Agent Pay and Visa's Trusted Agent Protocol use cryptographic signatures to let AI bots complete purchases securely. These agents analyze transaction patterns in real-time to detect fraud anomalies and adapt risk scoring automatically—meaning fewer false alarms for legitimate purchases.

AI Drug Discovery Startup ChemLex Raises $45M

AI-for-science company ChemLex announced major funding to expand its self-driving laboratory in Singapore. The autonomous chemistry system runs 24/7 experiments, capturing real-time data to accelerate pharmaceutical development. ChemLex already serves over 70 customers, including six of the world's top ten pharma companies. The AI-powered drug discovery market is projected to explode from $3.6B in 2024 to nearly $50B by 2034.

Tokyo Electron Shifts AI Focus

Tokyo Electron projects AI-driven semiconductor sales will hit 40% of revenue by FY2026, offsetting China market slowdown. This acceleration shows manufacturing equipment makers are betting heavily on AI infrastructure buildout.

Rubrik Launches AI Agent Protection Tools, Stock Surges 24.5%

Rubrik announced new AI governance offerings designed to monitor and protect AI agents across cloud platforms. The company launched Rubrik Agent Cloud for Amazon Bedrock AgentCore, enabling businesses to track and control AI agents in real-time.

Why This Matters: As AI agents become standard in business operations, security is crucial. This tool lets companies "rewind" and audit what their AI agents do, preventing errors or misuse.

Key Results: Q3 fiscal 2026 showed $350.17 million revenue and reduced net losses of $63.83 million. The stock jumped 24.5% on the announcement, signaling investor confidence.

Practical Action: If your company plans to deploy AI agents in 2026, you now have governance solutions available. Rubrik positions itself at the intersection of AI safety and data protection—exactly what enterprises need as they scale autonomous AI systems.

The Takeaway: AI agent management isn't theoretical anymore. Tools exist now to govern them safely across multi-cloud environments. Companies deploying agents without oversight may face security and compliance risks.