Oracle Updates Autonomous AI Lakehouse for Live Data Analysis

Key Takeaways

• <ul><li>Oracle's Autonomous AI Lakehouse updates let AI agents analyze live data across mixed systems without moving it</li><li>A new Business Semantic Layer helps AI agents understand KPIs and business metrics</li><li>Oracle's Business Simulation Engine supports scenario planning and forecasting within 12 months</li><li>Analysts view Oracle's approach as strong for the agentic AI market</li><li>Anthropic seeks $3.7 billion in its IPO, backed by Google and Amazon, with $1.4 billion revenue and $1.1 billion loss</li><li>Global agencies urge patching of critical PHP vulnerabilities affecting 70% of websites</li><li>Meta adds visual data opt-out for Ray-Ban glasses but US voice data rules stay unchanged</li><li>Credit default swaps tied to Oracle, Meta, Nvidia and others averaged over $9 billion weekly in September</li><li>AI factories produce 20-30 times more heat per unit area than commercial buildings, Johnson Controls reports</li><li>50 Korean firms use AI synthetic shoppers, with 91.2% match to human shopper surveys</li></ul>

Oracle brings AI agents to live enterprise data with Autonomous AI Lakehouse updates

Oracle announced new Autonomous AI Lakehouse features that let AI agents understand live data across Oracle and non-Oracle systems. The updates include an AI Data Catalog, a Business Semantic Layer, and tools for running business simulations. Data stays in its original location while AI and users analyze it in place. The goal is to reduce data copying, lower costs, and keep security and governance consistent across the enterprise.

Analysts praise Oracle Autonomous AI Lakehouse for business-aware AI

Industry analysts reacted positively to Oracle's updated Autonomous AI Lakehouse. They highlighted its ability to bring AI agents directly to live data across Oracle, non-Oracle, and Apache Iceberg sources. Analysts praised the Business Semantic Layer, multidimensional simulations, and strong governance controls. They say the approach reduces data movement, improves decision-making, and positions Oracle strongly in the agentic AI market.

Oracle launches Deep Trust AI Architecture to secure AI applications

Oracle announced its Deep Trust AI Architecture, a blueprint for embedding security into enterprise AI systems. The architecture covers software platform, data, data access, and application development layers. It includes automated patching, database-enforced access rules, and controlled data access for AI agents. Oracle offers security and patching tools free or at a discount through February 2027 for cloud co-managed and on-premises systems.

Oracle expands Deep Data Security for analytics, APEX, ORDS, and AI agents

Oracle released AI Database Update 23.26.4 to extend Deep Data Security across more enterprise tools. The update adds data grant filters that restrict access based on a user's business scope. It also supports analytics dashboards, APEX applications, ORDS REST APIs, and AI agents that generate SQL dynamically. The system enforces authorization at the database level so users and agents cannot bypass security rules.

Oracle details Deep Trust AI Architecture for trusted enterprise AI

Oracle's Deep Trust AI Architecture provides a practical blueprint for safer enterprise AI use. It addresses risks across four layers: software platform, data, data access, and application development. The design emphasizes enforcing security and business rules at the data source, not just in application code. It also highlights the need to protect data from AI agents that may access databases directly and bypass application-level protections.

Oracle launches semantic layer and simulation tools for AI Lakehouse

Oracle is adding a Business Semantic Layer to its Autonomous AI Lakehouse to help AI agents understand business metrics and KPIs. The company also launched a Business Simulation Engine for scenario planning and forecasting. Oracle is expanding support for Apache Iceberg to simplify data management. These tools will be available within the next 12 months. The new features require enterprises to manage data governance carefully.

Anthropic IPO filing highlights power and climate risks for AI data centers

Anthropic warned in its IPO filing that opposition to its data centers and rising power and water demand could raise costs. The company also cited climate change and tighter regulation as risks to its reputation and operations. Anthropic is seeking to raise $3.7 billion in its IPO. The company is backed by Google and Amazon. The filing draws attention to the environmental impact of the AI industry.

Anthropic IPO reveals $1.4 billion revenue and $1.1 billion loss

Anthropic filed its S-1 with the SEC, showing $1.4 billion in revenue last year and a net loss of $1.1 billion. The losses are driven by heavy investment in research and development. The company is seeking $3.7 billion in its IPO, which is expected in the coming weeks. The prospectus also warns that AI could be used for harmful purposes like misinformation. Major investors include Google and Amazon.

Global agencies urge patching of critical PHP vulnerabilities

Global cybersecurity agencies urge organizations to patch two critical PHP vulnerabilities known as PHP2Shell. The flaws, tracked as CVE-2026-109618 and CVE-2026-109619, could allow attackers to run code remotely without access. Australian government websites were taken offline as a precaution. CISA recommends updating to PHP 8.5.12, 8.4.27, 8.3.36, or 8.2.35. Researchers reported that AI helped identify how the vulnerabilities could be exploited.

PHP security flaws affect most of the internet and prompt shutdowns

Critical vulnerabilities in PHP, the language behind about 70% of websites, could let attackers control servers without passwords. The PHP Foundation admitted the flaws were known since at least 2024 and fixing them was delayed. Government websites in the UK, Finland, and Japan were disrupted. AI agents helped researchers discover and test fixes. The FBI's cyber division also warned about the risks.

Deep reinforcement learning improves stock trading with sentiment analysis

Researchers proposed a deep reinforcement learning approach for stock trading that uses sentiment analysis. The method combines actor-critic agents with Q-learning principles like DDPG, TD3 and SAC. Tests showed that policy optimization combined with Q-learning improved performance in profit maximization and risk-averse trading when deep sentiment data was infused.

AI factories face rising heat and energy demands, Johnson Controls says

AI factories now produce 20 to 30 times more heat per unit area than commercial buildings. Up to 40 percent of data center power goes to cooling. Johnson Controls released a reference design that integrates power generation, cooling and energy optimization. The design lowers PUE to 1.23 and can cut cooling carbon emissions by up to 43 percent.

Frenos report shows AI attack planning still needs human oversight

Frenos tested leading AI models on tasks related to industrial control system security. The models produced passing attack plans only about two-thirds of the time and often skipped steps. Frenos recorded 16,980 tests across decision tasks, planning tasks and simulated scenarios. The report stresses that AI developers and defenders must tightly control access to frontier models.

Meta adds visual data opt-out for Ray-Ban glasses, US voice rules unchanged

Meta now lets Ray-Ban smart glasses owners stop storing visual data from AI requests through a toggle in the Meta AI app. The setting blocks future storage, training and human review of that visual data. US users still cannot opt out of default voice data storage, while UK and German users have more choices. European regulators continue to raise privacy concerns.

Investors increase CDS trading as AI-related debt surges

Credit default swaps tied to major tech companies averaged over $9 billion per week in September, up from under $2 billion in July. Morgan Stanley estimates AI-linked companies issued about $466 billion in debt in 2026, compared with $101 billion the prior year. Companies like Oracle, Meta, Nvidia and SpaceX saw record CDS spreads as investors sought protection from credit risk.

SHFE and LME tin prices rise on low inventory and slower AI trading

The most-traded SHFE tin contract closed at 385,250 in the night session, up 1.81%. LME tin 3M closed at 50,765, up 1.35%. The gains were supported by low inventory and cooling AI trading. The report came from SMM Tin Morning Meeting Summary.

50 Korean firms use AI shoppers to test products

South Korean retail and consumer brands are using AI-generated synthetic shoppers to test products, set shelf prices and plan store layouts. Technology vendor Intellicia has finished 120 proof-of-concept projects across 50 companies. It reported a 91.2 percent match with a survey of 500 human shoppers. Lotte Members is building an AI Twin Consumer using data from 45 million loyalty members. Coupang, Naver and BGF Retail are also using synthetic buyers to test online and in-store shopping experiences.

Sources

• Activate AI Agents on your entire enterprise estate with innovations in Oracle Autonomous AI Lakehouse
• Analyst perspectives on new Oracle Autonomous AI Lakehouse capabilities
• Oracle says new AI apps inherit security protections from the system underneath them
• Oracle Deep Data Security Expands Authorization Across Analytics, APEX, ORDS, and AI Agents
• Announcing: Oracle Deep Trust AI architecture
• Oracle adds a semantic layer to ground AI agents in the business context
• Anthropic IPO filing warns of power, water and climate-related risks
• Anthropic’s IPO Prospectus Reveals What’s Under the Hood
• Critical PHP vulnerabilities raise worldwide concerns as government websites go offline, researchers report AI-assisted attacks
• What to know about the AI-detected security bug impacting wide swaths of the internet
• Financial Context-Rich Strategy for Actor-Critic based Agents: A Deep Reinforcement Learning Approach for Stock Trading
• With the computing density of artificial intelligence (AI) factories rising to unprecedented levels,..
• Frenos warns advances in AI attack planning could increase cybersecurity risks for critical infrastructure, ICS systems
• Meta Lets Ray-Ban Wearers Block AI Training on Camera Data, but US Users Still Can't Stop Voice Storage
• Are Investors Increasing Protection Against AI Debt Risk? CDS Trading Volumes Rise
• The most-traded SHFE tin contract closed at 385,250 in the night session, up 1.81%, while LME tin 3M closed at 50,765, up 1.35%, supported by low inventory and cooling AI trading [SMM Tin Morning Meeting Summary]
• 50 Korean Firms Turn to Synthetic AI Shoppers to Test Products