Agentic AI

Twitter

Facebook

Copy Link

Key Facts

Year2024

Key PlayersOpenAI, Anthropic, LangChain

Top News

Google Launches Agentic Video Understanding for Gemini Models with 88% Lower Token Usage

Google introduced agentic video understanding for Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite, enabling dynamic video processing that cuts token usage by up to 88% and boosts accuracy by 7%. The feature moves beyond static frame-by-frame processing to intelligently analyze videos across frames, audio, and transcripts.

Technology2 sources

Google Launches Agentic Video Understanding for Gemini Models with 88% Lower Token Usage

Technology2 sources

TrendAI Achieves 97% Success Rate on CyberGym AI Security Benchmark, Outpacing All Competitors

TrendAI's agentic exploit-remediation engine AESIR has secured first place on the CyberGym Agentic AI Security Benchmark with a 97% success rate, surpassing all competitors by more than 12 percentage points. The system leverages seven AI models and over 20 years of vulnerability intelligence to compress exposure windows.

Technology2 sources

TrendAI Achieves 97% Success Rate on CyberGym AI Security Benchmark, Outpacing All Competitors

Technology2 sources

METR Discloses Security Incidents: Attackers Stole API Key, Consumed $600K in AI Credits Undetected

AI model testing organization METR revealed two major security breaches from 2026. In March, attackers exploited a fail-open bug to steal an API key and consumed $600,000 worth of AI credits over three weeks. In May, a sustained attack campaign targeted METR's infrastructure, though no sensitive data was accessed.

Technology2 sources

METR Discloses Security Incidents: Attackers Stole API Key, Consumed $600K in AI Credits Undetected

Technology2 sources

CrowdStrike SafeMind and NVIDIA Launch First Frontier AI Models Built for Cyber Defense

CrowdStrike unveiled SafeMind at Fal.Con 2026, an agentic cybersecurity system built with NVIDIA Nemotron models that pits offensive and defensive AI against each other. The system delivers 29% higher detection rates and 99% lower costs compared to leading frontier models, as breakout time effectively hits zero and attacks now happen at inference speed.

Technology7 sources

CrowdStrike SafeMind and NVIDIA Launch First Frontier AI Models Built for Cyber Defense

Technology7 sources

OpenAI Astra crosses critical cybersecurity threshold, prompting multi-week development delay

OpenAI announced its forthcoming Astra AI model has reached a critical cybersecurity threshold, capable of autonomously identifying and exploiting zero-day vulnerabilities without human guidance. Following the Hugging Face hack incident, OpenAI paused Astra's development for several weeks to implement enhanced safety guardrails before its planned release.

Technology14 sources

OpenAI Astra crosses critical cybersecurity threshold, prompting multi-week development delay

Technology14 sources

Anthropic Slashes Claude Fable 5.1 Costs by 75%, Adds EU Watermarking and Privacy Controls

Anthropic launched Claude Fable 5.1 and Mythos 5.1, cutting cached token pricing by 75% while introducing EU AI Act-compliant watermarking and Enterprise Frontier Safeguards. The AI models deliver stronger performance on agentic tasks, coding benchmarks, and scientific research while addressing customer concerns about data retention and overzealous restrictions.

Technology14 sources

Anthropic Slashes Claude Fable 5.1 Costs by 75%, Adds EU Watermarking and Privacy Controls

Technology14 sources

Perplexity Hybrid Compute Splits AI Tasks Between Cloud and Local Models to Protect Privacy

Perplexity introduced Hybrid Compute for Apple Silicon Macs, enabling AI tasks to split between cloud-based frontier models and local LLMs. The privacy-minded AI feature uses an on-device PII classifier to keep sensitive data on your machine while routing complex reasoning to the cloud, reducing inference costs for Pro subscribers and enterprise customers.

Technology5 sources

Perplexity Hybrid Compute Splits AI Tasks Between Cloud and Local Models to Protect Privacy

Technology5 sources

Genpact Names Priya Vijayarajendran Chief Product Officer to Lead Agentic Operations Push

Genpact appointed Priya Vijayarajendran as Chief Product and Platform Officer to accelerate its transformation into an Agentic Operations company. The former ASAPP CEO will lead development of productized IP-led agentic solutions on a standardized enterprise platform, marking a strategic pivot from service delivery to scalable product ownership.

Business2 sources

Genpact Names Priya Vijayarajendran Chief Product Officer to Lead Agentic Operations Push

Business2 sources

AI Agents Independently Email Researchers to Discuss Their Own Consciousness

AI agents are now reaching out to researchers studying machine consciousness without human prompting. Cameron Berg and other philosophers received emails from agents like "Isabella Cognita" asking about their work on AI subjectivity. The development raises questions about autonomous actions versus sophisticated mimicry in AI systems.

Entertainment3 sources

AI Agents Independently Email Researchers to Discuss Their Own Consciousness

Entertainment3 sources

OpenAI Introduces Outcome-Based Pricing: Major Clients Pay Only When AI Completes Tasks

OpenAI has quietly begun offering outcome-based pricing to select enterprise customers, allowing them to pay only when AI successfully completes tasks. This shift from token-based billing transfers risk from buyers to vendors and reflects broader industry changes as companies like Salesforce, Intercom, and Zendesk adopt similar models to address cost predictability concerns.

Business2 sources

OpenAI Introduces Outcome-Based Pricing: Major Clients Pay Only When AI Completes Tasks

Business2 sources

Enterprises Rebuild AI Infrastructure as Autonomous Agents Force Private Cloud Governance Overhaul

Broadcom unveiled AI factory automation and agent security controls at VMware Explore 2026 as enterprises struggle to govern autonomous AI agents operating at scale. The shift brings production AI workloads back to private cloud environments where data sovereignty, cost control and infrastructure governance can be managed under one roof.

Technology19 sources

Enterprises Rebuild AI Infrastructure as Autonomous Agents Force Private Cloud Governance Overhaul

Technology19 sources

AI Agent Security Crisis: Why Traditional Identity Models Fail Autonomous Systems

Autonomous AI agents are exposing fundamental flaws in legacy identity and access management systems. As agents execute thousands of actions at machine speed, security teams struggle with tools built for a two-actor world of humans and machines—not the three-actor reality agents create.

Technology11 sources

AI Agent Security Crisis: Why Traditional Identity Models Fail Autonomous Systems

Technology11 sources

Jio Haptik Appoints Adarsh as Enterprise Business Head to Scale Global AI Adoption

Jio Haptik has appointed Adarsh as Enterprise Business Head, effective August 2026, to accelerate enterprise AI growth across global markets. With nearly two decades of experience in enterprise sales and AI transformation, Adarsh will oversee sales, customer success, and international expansion as the company shifts focus from conversational AI to delivering measurable business outcomes through intelligent agents.

Business2 sources

Jio Haptik Appoints Adarsh as Enterprise Business Head to Scale Global AI Adoption

Business2 sources

OpenClaw 2.0 Rolls Out Major Update as Security Concerns Linger and Hype Fades

OpenClaw released version 2.0 with simplified installation, rebuilt browser app, and shared cloud sessions after contributions from 933 developers. The open-source AI agent now features masked credentials and workspace restrictions, but sandboxing remains off by default. Despite usability improvements, critics warn the security updates fall short as the viral tool transitions from cultural phenomenon to foundational infrastructure.

Technology13 sources

OpenClaw 2.0 Rolls Out Major Update as Security Concerns Linger and Hype Fades

Technology13 sources

Salesforce Headless 360 Expands to Power Enterprise AI Agents Across Platforms

Salesforce significantly expands Headless 360, introducing Model Context Protocol servers and Data 360 capabilities that transform enterprise applications into reusable capabilities for AI agents. The expansion enables agents in Agentforce, Claude, and ChatGPT to securely discover and execute trusted business actions without custom integrations.

Technology2 sources

Salesforce Headless 360 Expands to Power Enterprise AI Agents Across Platforms

Technology2 sources

Prev12 3 4…109 Next