Okta Expands Agent Management, Cautions on AI Speed

Key Takeaways

• Okta launched Agent SSO to allow businesses to add AI agent identities to directories without extra charges.
• Some Okta customers saw their agent numbers increase from 50 to 1,500 in just a few weeks.
• Okta President Eric Kelleher presented a Blueprint for the Secure Agentic Enterprise at the Goldman Sachs conference.
• Anthropic CEO Dario Amodei called for slowing down the rapid pace of AI development to manage risks.
• OpenAI, Sam Altman, and Elon Musk support a coordinated effort to slow AI development using third-party safety evaluators.
• AI has created approximately one million new US jobs since mid-2023, driven by data center construction.
• Approximately 200,000 US jobs were linked to AI layoffs during the same period, far fewer than new creations.
• CrowdStrike launched Project QuiltWorks for North America to help enterprises remediate AI security issues.
• CrowdStrike has a market cap of $213.9 billion as it expands its AI risk services.
• Ilit Raz pleaded guilty to securities fraud, defrauding investors of approximately $27 million through false claims about Joonko.

Okta Helps Companies Secure Confusing AI Agents

Okta President and COO Eric Kelleher spoke at the Goldman Sachs Communacopia + Technology conference about the need to secure AI agents. He noted that many enterprise customers feel confused about how to protect these new technologies. Okta is addressing this by offering a Blueprint for the Secure Agentic Enterprise that focuses on discovery, connectivity, and authorization. The company recently launched Agent SSO, which allows businesses to add agent identities to their directories without extra charges. Early sales show that some customers saw their agent numbers jump from 50 to 1,500 in just a few weeks. Okta plans to price its service as an add-on to existing user fees to make it easier for companies to start.

Okta Expands Identity Platform to Manage AI Agents

Okta is expanding its Universal Directory to include identities for AI agents as part of its new offering. Eric Kelleher explained that the company positions its solution around three key questions: where agents are deployed, what systems they connect to, and what actions they can perform. The new Agent SSO tool is now generally available on the workforce platform and helps companies track and govern these agents just like human employees. Okta completed dozens of transactions since the product launched on April 30, including a large deal with a Fortune 50 healthcare company. To reduce friction, the company initially charges based on the number of users rather than the number of agents. Kelleher believes this approach helps customers who are unsure about how many agents they will need.

Okta Addresses Security Challenges for Growing AI Agents

Okta President and COO Eric Kelleher highlighted that uncertainty around AI agents is a major challenge for the identity-security market. Speaking at the Goldman Sachs Communacopia + Technology conference, he stated that customers need clear guidance on how to secure these autonomous software tools. Okta sees a significant opportunity to help organizations manage access to their systems and data as AI adoption grows. The company is working closely with clients to understand specific risks and develop strategies to mitigate them. Kelleher emphasized that every agent requires a trackable identity, even if some are temporary. He expressed excitement about the growing interest in Okta's platform from companies looking to protect their AI investments.

Anthropic CEO Calls for Slower AI Development Pace

Anthropic CEO Dario Amodei has publicly stated that the world should slow down the rapid advance of artificial intelligence. He believes the current speed of development poses significant risks that need to be managed carefully. This call for caution comes as AI technology continues to evolve quickly across the industry. Amodei's comments suggest that a more measured approach is necessary to ensure safety and stability. The article notes that this perspective is shared by other major figures in the field who are concerned about the pace of progress.

Top AI Leaders Support Call to Slow Down Development

Leading figures in the artificial intelligence industry are supporting a coordinated effort to slow down AI development. Anthropic CEO Dario Amodei proposed this idea and suggested using third-party safety evaluators to oversee the process. Tech giants like OpenAI and figures such as Sam Altman and Elon Musk have rallied around this call for caution. The group believes that slowing the pace will help address safety concerns before they become critical issues. This unified stance shows that major players in the sector agree that current speed is too fast for safe deployment.

AI created one million new US jobs

Artificial intelligence has created about one million new jobs in the United States since mid-2023. This number is roughly five times higher than the approximately 200,000 layoffs linked to the technology during the same period. The growth is driven by the construction of data centers and new power generation facilities. These projects have increased demand for electricians, HVAC technicians, grid engineers, and service technicians. Some occupations like paralegals have also seen growth, with numbers rising by about 11 percent between 2023 and 2025.

CryptoL framework improves financial forecasting

Researchers introduced CryptoL, a new framework designed to improve cryptocurrency forecasting using multivariate time series data. The tool addresses challenges like extreme differences in asset scales and non-stationary dynamics within Open, High, Low, and Close variables. CryptoL uses a RevIN pipeline to evaluate errors in context-normalized coordinates, which prevents distortion from inverse normalization. It also includes scale-adaptive numerical stabilization and a soft feasibility loss to ensure predictions respect OHLC inequalities. Tests across various cryptocurrencies show the framework improves accuracy, training stability, and the frequency of valid financial predictions.

Joonko founder pleads guilty to fraud

Ilit Raz, the founder and former CEO of Joonko Diversity Inc, pleaded guilty to securities fraud on September 12, 2026. She admitted to a scheme that defrauded investors out of approximately 27 million dollars by lying about the company's customers and revenue. Joonko claimed to offer an AI product to help employers hire diverse candidates, but Raz fabricated documents to support these false claims. The case was assigned to U.S. District Judge Alvin K. Hellerstein and handled by the Southern District of New York. The company is now bankrupt, leaving many investors with significant financial losses.

Experts urge physical safeguards for AI safety

An opinion piece argues that relying on treaties alone is not enough to prevent dangerous artificial intelligence systems. The author suggests that the world should consider not building certain types of autonomous robots known as murderbots. Artificial intelligence safety is currently viewed as the most critical policy issue facing the globe. The article highlights the need for stronger physical safeguards to avoid potential mass disasters caused by rogue AI systems.

CrowdStrike expands AI risk services in North America

CrowdStrike Holdings has launched a localized version of its Project QuiltWorks program for North America. This initiative connects US and Canadian enterprises with partners like Blackthorne Consulting, CDW, GuidePoint Security, and Presidio to provide AI risk services. The program combines frontier AI models, cloud platforms, and cyber insurance to help companies remediate security issues. CrowdStrike aims to use its Falcon Flex and AI-driven tools to deepen customer commitment across cloud, identity, and SIEM systems. With a market cap of 213.9 billion dollars, the company positions this move as a key part of its strategy to secure critical systems using specialized AI support.

OpenAI and Samsung partner with Broadcom on new AI chips

OpenAI and Samsung are working together to create new AI chips designed for large business operations. This partnership builds on earlier joint processor work between OpenAI and Samsung. Broadcom plays a key role in this project by supplying high performance chips needed for AI workloads. Samsung plans to use these custom processors in its own critical AI tasks. This move strengthens Broadcom's position in the custom AI hardware market while expanding OpenAI's hardware strategy.

Vontier stock rises with new AI charging software launch

Vontier recently launched an AI powered Network Optimization Agent through its subsidiary Driivz at the ICNC26 event. This new tool focuses on electric vehicle charging software and data driven operations. The company stock closed at $32.17 after a recent 2.91% gain, though it remains down 14.89% for the year so far. Investors are watching to see if this AI charging push can improve the company's performance. Vontier inherited strong management practices from its former parent companies Danaher and Fortive.

Zscaler CEO says AI security demand is growing fast

Zscaler CEO Jay Chaudhry announced at the Citi TMT Conference that demand for AI security solutions is increasing rapidly. He stated that customers are adopting zero-trust architecture to protect their networks from cyber threats. Zscaler's new solutions use machine learning to detect and stop threats in real time. The company aims to work well with existing security systems while continuing to innovate. Chaudhry believes their technology will remain critical for securing networks against future cyber attacks.

Sources

• Okta Targets AI Agent Confusion With Identity Security Push
• Okta Targets AI Agent Confusion With Identity Security Push
• Okta Targets AI Agent Confusion With Identity Security Push
• Dario Amodei says he's now 'convinced' world should slow AI's advance
• Amodei, Altman, and Musk rally around call to slow AI down
• 📊 AI has created a million new jobs in the US
• CryptoL: Towards Scale Dominance and Physics Constraints Mitigation in Financial Multivariate Time Series Forecasting
• Founder And Former CEO Of Artificial Intelligence Company Pleads Guilty To Securities Fraud
• Don’t rely on treaties to stop the murderbots
• What Does CrowdStrike (CRWD) North America QuiltWorks Rollout Mean For AI Risk?
• What Does Broadcom (AVGO) Gain As Its AI Chip Ties Deepen?
• Vontier (VNT) Draws Attention With AI Charging Push, Is It Still Below Fair Value?
• Zscaler CEO Highlights AI Security Surge, Zero-Trust Momentum at Citi Conference