September 27, 2025

news  
openai, databricks and meta Updates

The rapid advancement and integration of artificial intelligence are reshaping industries and raising critical questions about security, ethics, and the future of human intelligence. OpenAI CEO Sam Altman predicts AI could surpass human intelligence by 2030, with AI potentially handling a significant portion of economic tasks. This surge in AI capabilities is driving partnerships, such as the one between OpenAI and Databricks, to enable enterprises to build AI applications using their own data with enhanced governance. Meanwhile, companies like Okta are addressing the security challenges posed by autonomous AI agents, launching an Identity Security Fabric to manage their lifecycles and access across applications, a move supported by major tech firms and aligned with zero-trust principles. The European Commission is also scrutinizing Big Tech firms like Meta and Google to ensure their AI integrations comply with digital competition rules and data usage obligations. Beyond enterprise applications, AI is being explored for societal benefits, from protecting authentic goods in Vietnam's online market using AI-powered labels to teaching AI literacy in Beijing schools to cultivate future innovators. State governments are cautiously adopting AI, primarily in low-risk areas, while public skepticism remains. Ethical considerations are also at the forefront, with discussions at the University of Notre Dame exploring AI's impact on faith and humanity through a new DELTA framework, and UA Little Rock hosting panels on AI ethics, data privacy, and manipulation.

Key Takeaways

  • Okta introduces Identity Security Fabric to manage AI agent lifecycles and access, aiming to unify scattered security solutions.
  • Zero-trust security principles are crucial for managing autonomous AI agents with broad access, addressing risks from insecure practices like hard-coded passwords.
  • OpenAI CEO Sam Altman forecasts AI surpassing human intelligence by 2030 and handling 30-40% of economic tasks.
  • OpenAI and Databricks are partnering to allow enterprises to build AI applications and agents using their own data on the Databricks Data Intelligence Platform.
  • The European Commission is monitoring Meta and Google to ensure AI integration complies with digital competition rules and data use obligations under the DMA.
  • AI is being used to combat counterfeit goods in Vietnam's online market through AI-powered labels that provide product traceability.
  • State governments are cautiously adopting AI, with most deploying it in low-risk areas like chatbots, while public trust in government AI services remains low.
  • Beijing is implementing AI literacy classes in primary and secondary schools to foster future innovators.
  • A new framework called DELTA (Dignity, Embodiment, Love, Transcendence, Agency) is proposed to guide ethical discussions on AI's impact on faith and humanity.
  • Sakana AI's open-source ShinkaEvolve framework uses LLMs to evolve programs efficiently for scientific discovery with fewer evaluations.

Okta launches Identity Security Fabric for AI agents

Okta introduced its new Identity Security Fabric to help businesses secure AI agents. This platform manages AI agent lifecycles, controls access across applications, and uses verifiable credentials to reduce risks. It aims to replace scattered security solutions with a unified system. The fabric will help organizations manage AI systems that often have high access levels but lack oversight. Gartner predicts identity fabric principles will prevent 85% of new attacks by 2027.

Zero-trust security is key for managing AI agents

Zero-trust security is becoming essential for managing AI agents, which have autonomy, broad access, and operate at scale. Okta's chief security officer, David Bradbury, emphasizes that zero-trust principles like secure identity, least privilege, and continuous monitoring are critical for these non-human identities. Companies are rushing to adopt AI, sometimes cutting security corners by using hard-coded passwords or outdated API keys. This can lead to significant risks, especially with the proliferation of access tokens across many AI agents and services.

Okta integrates AI agents into security fabric

Okta is introducing new features to integrate AI agents into its identity security fabric, addressing the security gap as 91% of organizations use AI agents but only 10% have management strategies. The 'Okta for AI Agents' component helps identify risks, manage access, and govern AI agent activity. Cross App Access, an extension of OAuth, secures communication between AI agents and applications, supported by major tech companies. Verifiable Digital Credentials (VDC) will be available in 2027 to establish trust and combat AI fraud.

Sam Altman predicts AI will be smarter than humans by 2030

OpenAI CEO Sam Altman believes artificial intelligence will surpass human intelligence by 2030, stating that current models are already smarter than him in many ways. He predicts AI could handle 30% to 40% of economic tasks in the near future. Other AI leaders like Dario Amodei and Elon Musk anticipate this milestone even sooner. Altman also noted that human connection and understanding what others want will remain important. OpenAI is building significant data center infrastructure to support AI advancements.

AI helps protect authentic goods in Vietnam's online market

Vietnam's e-commerce platforms face a growing problem with counterfeit goods, with authorities removing thousands of products and sanctioning online shops. To combat this, businesses are using technology like AI to protect their brands and consumers. Companies are developing AI-powered labels that act as digital passports, providing traceability and instant verified information about product origin and quality. This technology helps build consumer trust and offers a proactive defense against counterfeiting in the digital marketplace.

States' cautious AI adoption faces challenges

A new report from the National Association of State Chief Information Officers (NASCIO) indicates that state governments are cautiously adopting AI, but this approach may become harder to justify as AI capabilities improve. While many government employees believe AI will enhance citizen experience and reduce workloads, the public remains skeptical about AI in government services. States are currently deploying AI in low-risk areas like chatbots and FAQs, with only 6% having fully scaled AI across agencies. The report recommends strengthening governance and prioritizing user needs.

OpenAI and Databricks partner for enterprise AI agents

OpenAI and Databricks have joined forces to help businesses build AI applications and agents using their own data. This partnership makes OpenAI's models available on the Databricks Data Intelligence Platform and Agent Bricks. Enterprises can now develop AI solutions with strong governance and performance, leveraging their secure data. Both companies aim to continuously improve AI models for business needs, simplifying the deployment of advanced AI for various industry uses like customer service, healthcare, and finance.

Notre Dame summit explores AI's impact on faith and humanity

The University of Notre Dame hosted a summit exploring the ethical and moral questions surrounding artificial intelligence. Educators, faith leaders, policymakers, and technologists discussed AI's impact on the human experience from a Christian perspective. The event launched a new framework called DELTA (Dignity, Embodiment, Love, Transcendence, Agency) to guide conversations about AI. Speakers emphasized the importance of human dignity, real connections, and moral responsibility in the age of AI, urging faith-based perspectives to be included in these crucial discussions.

Sakana AI's ShinkaEvolve evolves programs efficiently

Sakana AI has released ShinkaEvolve, an open-source framework that uses large language models to evolve programs for scientific discovery with significantly fewer evaluations. The system combines adaptive parent sampling, novelty-based filtering, and an adaptive LLM ensemble to reduce the need for extensive testing. ShinkaEvolve sets a new standard in circle packing with only about 150 evaluations, outperforms other systems in math reasoning and competitive programming, and discovers a new load-balancing loss for AI training. The code and research report are publicly available.

Beijing schools teach AI literacy to young innovators

Beijing is introducing AI literacy classes in primary and secondary schools this semester to help students understand and develop skills in artificial intelligence. The initiative aims to nurture future innovators by teaching students about AI technology. The program focuses on shaping students' views of technology and preparing them for an increasingly AI-driven world.

EU warns Big Tech on AI integration and data use

The European Commission is closely monitoring how major tech companies integrate AI into their services to ensure compliance with digital competition rules. Commissioner Henna Virkkunen stated that platforms like Meta and Google must adhere to obligations regarding data processing and cross-use between services, especially for designated gatekeepers under the Digital Markets Act (DMA). The Commission is also examining how AI rules can foster a competitive AI sector in the EU, emphasizing that AI integration should not extend dominant positions into the AI market unfairly.

UA Little Rock hosts AI ethics panel discussion

UA Little Rock Downtown will host a public panel discussion on the ethical implications of artificial intelligence on October 9th. Experts from computer science, philosophy, history, and information science will explore questions about using AI ethically and the risks of its rapid spread. The panel aims to spark community conversation about AI's impact on data privacy, user manipulation, and environmental costs, encouraging responsible adoption of the technology.

Sources

AI agents Identity security Zero-trust security Access management Verifiable credentials AI governance AI ethics AI adoption Enterprise AI AI literacy AI regulation AI fraud Counterfeit goods Brand protection AI development AI models Data privacy Digital transformation AI security AI applications