Exclusive: WitnessAI nabs $58M to secure enterprise AI axios.com/2026/01/13/witness…

WitnessAI Secures $58M to Grow Global AI Security Reach bankinfosecurity.com/witness…

TENEX.AI and WitnessAI Partner to Deliver Managed Security and Compliance Services for Enterprise AI - businesswire.com/news/home/2…

WitnessAI Achieves SOC 2 Type II Compliance, Reinforcing Commitment to Security and Trust prnewswire.com/news-releases…

RSAC2026フル参戦が決まりまして先ほどフライトを確保しました。現地時間、月曜日〜木曜日までフルでアテンドさせていただきます。ぜひ、WitnessAIブースでお待ちしております。 #RSAC2026 #WitnessAI @OneRSAC rsaconference.com/usa

30 Startups Hiring 👇 Higgsfield AI – generative media studio (Bay Area) WitnessAI – AI risk and governance (Bay Area / Atlanta / Remote) Railway – app deploy and hosting (Remote) LiveKit – live audio video APIs (US remote / Canada remote) Swap – commerce back office (London / Amsterdam / Austin / NYC) CHAMP – title transfer platform (US remote / Cleveland) Mesh – pay with crypto rails (Bay Area / NYC / Miami / US / EU) DealHub – sales quoting suite (Tel Aviv / Texas / Global) Datarails – finance planning tool (Tel Aviv / UK remote) Summize – in app contract flows (Boston) Ivo – contract analysis AI (Bay Area) Credit Key – net terms at checkout (US remote) Sokin – cross border money tools (London / NYC / Toronto) Propy Inc. – home sale closing tech (Miami / NYC / Sofia) Orbital – diligence automation (London / NYC) Cubby – storage operator OS (NYC / Lehi / Sofia) Snout – pet care memberships (Michigan / US remote) Tandem – faster Rx approvals (NYC) Stream – employee money perks (London / NYC) Duetti – music catalog buyer (NYC / LA / Miami) Superstate – digital asset manager (NYC) Jetson – home electrification installs (Colorado / NYC / Seattle / Vancouver) Hydrosat – farm insights from space (Luxembourg / US) Cyclic Materials – critical mineral recovery (Toronto / Kingston / Mesa) Northwood – satellite ground network (LA / DC) Freeform – automated metal printing (LA) RobCo – modular factory robots (Munich) Ethernovia – high speed networking (Pune / Bay Area / Canada) Aikido Security – security for dev teams (Ghent / Bay Area / US remote) osapiens – compliance and ESG stack (Munich / Madrid) Pomelo – cards and payments infra (Buenos Aires) Emergent – build apps with agents (Bay Area / Bangalore) Orbem – AI for imaging (Houston / Munich) Factify – docs to structured data (Tel Aviv) Wasabi Technologies – low cost object storage (US / Australia)

WitnessAI Launches Red-Teaming and AI Firewall Tools channelinsider.com/shared/ai… @Channel Insiderより

WitnessAIが提供するAI安全プラットフォーム、企業のデータ保護を強化 innovatopia.jp/ai/ai-news/27…

WitnessAI Raises $58 Million for Global Expansion and Announces New Ways to Secure AI Agents prnewswire.com/news-releases…

Linkedinでは告知させて頂きましたが、WitnessAI社に、 Senior Sales Engineer, EvangelistとしてJoinさせて頂きました。AIは使うのもだけど、使い方によっては悪影響を与えます、当社はそのAIを活用する上で必要なセキュリティ対策を具体的にご提案します。 witness.ai/ @witnessAI

WitnessAI raised $58M to build a security “confidence layer” for enterprise #AI, guarding against data leaks, compliance breaches and prompt-based attacks.🛡️📊 #AINews #AIInvestment #Cybersecurity #ShareForSuccess ow.ly/38iV50Y0zgn

📌 Q: What are the core components and processes of AI Threat modeling? There are 4 Core Components & Process to AI Threat Modeling 1. Understand the System: Map data flows, trust boundaries, and components (data, model, inference) 2. Identify What Can Go Wrong: Brainstorm AI-specific threats (e.g., data manipulation, prompt injection) 3. Mitigate Risks: Design defenses like data validation, input sanitization, and monitoring for anomalous outputs 4. Verify & Iterate: Confirm mitigations work and continuously update models as threats evolve v/ @witnessAI Cc: @EvanKirstel #AISecurity | #AIOps

WitnessAI Raises $58 Million For Global Expansion And Announces New Ways To Secure AI Agents Read more: financialit.net/news/fundrai… #funding #fundraising #WitnessAI #Cybersecurity #AIAgents #EnterpriseSecurity #TechExpansion #FintechInnovation #finance #fintech #FinancialIT

WitnessAI, which intercepts employees' custom GenAI models usage and applies safeguards, raised $58M led by Sound Ventures, taking its total funding to $85M (@ionutarghire / SecurityWeek) securityweek.com/witnessai-r… techmeme.com/260115/p18#a260… 📥 Send tips! techmeme.com/contact

Across the past two weeks, agentic AI continued shifting from assistants into tool-using systems that can complete workflows end to end, especially in commerce and enterprise operations. In parallel, enterprises and standards bodies pushed identity, authorization, observability, and governance layers to make autonomy safer and operable. Here’s your biweekly recap of everything that happened in the space, in the latest edition of AI Agents Pulse: 1️⃣ Google launched the Universal Commerce Protocol (UCP), an open standard for agentic commerce spanning discovery, checkout, and post-purchase support, aiming to reduce bespoke integrations across retailers and payment providers. blog.google/products/ads-com… 2️⃣ Google introduced Gemini Personal Intelligence (beta), letting the Gemini app connect to Gmail, Photos, Search, and YouTube so agents can reason over user context with opt-in controls and privacy guardrails. blog.google/innovation-and-a… 3️⃣ NVIDIA spotlighted Jetson Thor for reasoning-capable humanoids and released new physical-AI models and robotics tooling, signaling a push toward edge autonomy where agents perceive, plan, and act locally. investor.nvidia.com/news/pre… 4️⃣ Lenovo launched Lenovo Agentic AI and the xIQ suite, positioning a full-lifecycle enterprise stack to create, deploy, manage, and govern production-ready agents across hybrid environments with built-in automation. news.lenovo.com/pressroom/pr… 5️⃣ Microsoft Advertising unveiled Copilot Checkout and Brand Agents, enabling in-chat purchasing while merchants remain merchant-of-record, and letting Shopify stores deploy brand-voiced shopping agents with minimal integration. about.ads.microsoft.com/en/b… 6️⃣ Salesforce announced the Spring ’26 Release, expanding Agentforce-style capabilities across sales, service, data, and security workflows, framing an “agentic enterprise” model where human teams and agents co-run customer operations. salesforce.com/news/stories/… 7️⃣ Anthropic launched Cowork as a research preview in Claude Desktop, giving agents scoped access to folders and connectors for parallel task execution, while emphasizing explicit permissions and mitigation against prompt-injection risks. claude.com/blog/cowork-resea… 8️⃣ Snowflake announced intent to acquire Observe, aiming to unify logs, metrics, and traces with AI-powered observability built on open standards, and positioning telemetry as core infrastructure for operating production agents. snowflake.com/en/blog/observ… 9️⃣ CrowdStrike agreed to buy SGNL for $740M, adding continuous identity controls for humans, machines, and AI agents to support real-time authorization decisions as enterprises shift away from standing privileges. reuters.com/technology/crowd… 1️⃣0️⃣ WitnessAI raised $58M and announced expanded agentic AI governance, including visibility into which tools and MCP servers agents access, signaling growing demand for security and compliance layers around autonomous enterprise workflows. witness.ai/resources/witness… 1️⃣1️⃣ Deepgram raised $130M at a $1.3B valuation, highlighting sustained investment in voice infrastructure that powers real-time conversational agents across industries, plus continued M&A to accelerate speech-agent productization. reuters.com/technology/voice… 1️⃣2️⃣ NIST’s CAISI issued an RFI on securing AI agent systems, requesting input on safe development and deployment as agents combine models with tools that can change real-world state and create new attack surfaces. nist.gov/news-events/news/20… 1️⃣3️⃣ An IETF Internet-Draft proposed new authentication and authorization requirements for the AI agents era, focusing on dynamic delegation and ultra-low-overhead identity patterns better suited to high-frequency, ephemeral agent interactions. datatracker.ietf.org/doc/dra… 1️⃣4️⃣ Palantir published Ontology MCP, enabling developers to expose Foundry objects, actions, and queries as discoverable tools via Model Context Protocol, reducing bespoke integrations across agent frameworks while keeping access explicit. palantir.com/docs/foundry/an… 1️⃣5️⃣ Microsoft shipped a Visual Studio 2026 update with a built-in NuGet MCP server for GitHub Copilot, letting developers retrieve package intelligence and remediate known vulnerabilities through agent tooling directly inside the IDE. learn.microsoft.com/en-us/vi… 1️⃣6️⃣ Polygon Labs announced definitive agreements to acquire Coinme and Sequence, adding regulated fiat on/off-ramps and enterprise wallet infrastructure to pursue compliant stablecoin payments, a settlement layer that agentic commerce can program against. polygon.technology/blog/poly… 1️⃣7️⃣ Galaxy Research analyzed x402 and related agentic payment standards, arguing stablecoin-over-HTTP flows can let agents pay per request for APIs and data, making “machine commerce” viable without subscriptions or manual checkout. galaxy.com/insights/research… 1️⃣8️⃣Alibaba upgraded the Qwen app to natively connect Taobao, Alipay, Fliggy, and Amap, letting users complete food orders and travel bookings inside chat while authorizing in-conversation payments. reuters.com/world/china/alib…

$kite Over the past two weeks, Agentic AI has leveled up from chat assistants to full task-closing systems, especially in trading and enterprise workflows. Companies and standards groups are adding identity, authorization, observability, and governance layers to make autonomous agents safe and manageable in real-world production. Highlights: •Google launched UCP for seamless agentic commerce and Gemini Personal Intelligence to connect Gmail, Photos, Search, and YouTube with privacy safeguards •NVIDIA and Lenovo are pushing humanoid and enterprise-ready agents with local autonomy and lifecycle management •Microsoft, Salesforce, and Alibaba are integrating agents into commerce, sales, and everyday apps for real-world productivity •Anthropic, Snowflake, CrowdStrike, WitnessAI, and Deepgram are strengthening security, observability, identity, and voice infrastructure for reliable agent deployment •Standards bodies like NIST and IETF are defining safer, dynamic identity and authorization models for high-frequency agent interactions The future is clear: smart agents are moving from experiments to production, powering commerce, enterprise workflows, and everyday apps with safety, governance, and real-world impact @KiteAIChinese gkite @0xLaughing everybody.

在过去两周里，Agentic AI 继续从“会聊天的助手”走向“能调用工具并把任务闭环”的系统形态，最明显的落点集中在交易与企业工作流。与此同时，企业与标准组织也在加速补齐身份、授权、可观测与治理层，使自治能力在生产环境里更可控、更可运维。 以下是最新一期 AI Agents Pulse，对过去两周 Agentic AI 领域动态的回顾： 1️⃣ Google 发布 Universal Commerce Protocol（UCP）作为面向 agentic commerce 的开放标准，覆盖发现、结账与售后全链路，目标是减少商户与支付方之间为每个智能体单独对接的成本。 blog.google/products/ads-com… 2️⃣ Google 推出 Gemini「Personal Intelligence」（beta），让 Gemini App 在用户授权下连接 Gmail、Photos、Search 与 YouTube，使智能体能基于个人上下文推理，并提供清晰的开关与隐私护栏。 blog.google/innovation-and-a… 3️⃣ NVIDIA 聚焦 Jetson Thor 以支撑具备推理能力的人形机器人，并同步发布物理 AI 模型与机器人开发工具链，强化“边缘侧自治”路径，让智能体在本地完成感知、规划与行动。 investor.nvidia.com/news/pre… 4️⃣ 联想发布 Lenovo Agentic AI 与 xIQ 套件，主打覆盖创建、部署、管理与治理的企业级全生命周期栈，把生产就绪的智能体能力在混合环境中以自动化方式规模化落地。 news.lenovo.com/pressroom/pr… 5️⃣ Microsoft Advertising 发布 Copilot Checkout 与 Brand Agents，支持在对话内完成购买且商户仍为 merchant-of-record，并让 Shopify 商家以更低集成成本部署品牌口吻的购物智能体。 about.ads.microsoft.com/en/b… 6️⃣ Salesforce 发布 Spring ’26 版本更新，将 Agentforce 式能力扩展到销售、服务、数据与安全工作流，强化“Agentic Enterprise”叙事，即人类团队与智能体共同驱动客户运营。 salesforce.com/news/stories/… 7️⃣ Anthropic 以 research preview 形式发布 Claude Desktop 的 Cowork，为智能体提供对文件夹与连接器的受限访问并支持并行执行任务，同时强调明确授权与对 prompt injection 风险的防护。 claude.com/blog/cowork-resea… 8️⃣ Snowflake 宣布拟收购 Observe，目标是用 AI 可观测能力统一日志、指标与链路追踪并基于开放标准构建，强调遥测数据将成为生产级智能体可运维的核心基础设施。 snowflake.com/en/blog/observ… 9️⃣ CrowdStrike 宣布将以 7.4 亿美元收购身份安全公司 SGNL，引入 continuous identity 能力，为人、机器与 AI 智能体提供实时授权控制，帮助企业从长期静态权限迁移到动态访问决策。 reuters.com/technology/crowd… 1️⃣0️⃣ WitnessAI 完成 5800 万美元融资并发布更强的 agentic AI 治理能力，强调可见性与策略控制（包括智能体访问的工具与 MCP 服务器），反映企业对安全与合规层的需求持续升温。 witness.ai/resources/witness… 1️⃣1️⃣ Deepgram 完成 1.3 亿美元融资、估值达 13 亿美元，凸显语音基础设施仍是实时对话式智能体的关键底座，并通过扩张与并购推动语音智能体产品化加速。 reuters.com/technology/voice… 1️⃣2️⃣ NIST 旗下 CAISI 发布关于 AI 智能体系统安全的 RFI，征集业界与学界对安全开发与部署的建议，聚焦当智能体结合工具并能改变现实世界状态时产生的新攻防面。 nist.gov/news-events/news/20… 1️⃣3️⃣ IETF 发布面向 AI Agents 时代的认证与授权新需求草案，强调动态委托与更低开销的身份模式，以适配高频、短生命周期的智能体交互，而非仅依赖静态的人类中心身份假设。 datatracker.ietf.org/doc/dra… 1️⃣4️⃣ Palantir 发布 Ontology MCP，使开发者可通过 Model Context Protocol 将 Foundry 的对象、动作与查询暴露为可发现工具，降低不同智能体框架的定制对接成本并保持访问边界清晰可控。 palantir.com/docs/foundry/an… 1️⃣5️⃣ Microsoft 的 Visual Studio 2026 更新内置 NuGet MCP server 给 GitHub Copilot 使用，让开发者在 IDE 内通过智能体工具获取包情报并修复已知漏洞，推动 MCP 进入更主流的工程工作流。 learn.microsoft.com/en-us/vi… 1️⃣6️⃣ Polygon Labs 宣布签署收购 Coinme 与 Sequence 的最终协议，补齐受监管的法币出入金与企业钱包基础设施以推进合规稳定币支付，为 agentic commerce 提供可编程的结算轨道。 polygon.technology/blog/poly… 1️⃣7️⃣ Galaxy Research 深度解析 x402 与相关 agentic payment 标准，认为以 stablecoin-over-HTTP 实现“按请求付费”的 API/数据购买，将让机器经济在无需订阅与人工结账的条件下更可行。 galaxy.com/insights/research… 1️⃣8️⃣阿里升级千问 App，打通淘宝、支付宝、飞猪与高德等生态服务，让用户在对话内完成点外卖与出行预订，并通过对话内授权实现支付闭环。 reuters.com/world/china/alib…

WitnessAI Raises $58 Million for AI Security Platform securityweek.com/witnessai-r…

エンタープライズAIのセキュリティおよびガバナンスプラットフォームを提供するWitnessAIが5,800万ドルの資金調達を実施 #WitnessAI #ガバナンスプラットフォーム #エンタープライズAI startuplog.com/n/n05c7527e0c…

How WitnessAI raised $58M to solve enterprise AI’s biggest risk techcrunch.com/video/how-wit…