Nvidia Reports $115.8 Billion as AI Competition Intensifies

ServiceNow’s report on agentic security finds that 92% of Indian organisations experienced an AI-agent security incident in the past year, but only 36% contained one within 24 hours. Across the studies, just 19% had a named person responsible for AI security and 18% had a single platform to oversee agents; 37% found agents in production that security teams did not know about. While 51% monitored unusual activity in real time, 67% lacked automated response systems. The report surveyed 500 Indian security leaders, as part of a wider study of more than 3,700 decision-makers across 20 countries. ServiceNow’s Kulmeet Bawa called for agents to be registered, visible and governed with automatic controls and rapid response.

Nvidia’s quarterly revenue reached $115.8 billion, and its shares rose about 5% after hours. Its results, higher 2028 sales outlook and $150 billion buyback have renewed focus on spending for AI data centers, though leadership questions, investigations and export controls may influence where investment goes. The related articles examine potential beneficiaries among infrastructure companies: one identifies three stocks and a broader screen of 51 US and allied capacity builders; another finds 49 additional companies. The pieces say investors may see the developments as either an opportunity or a reason for caution.

Microsoft and OpenAI both introduce models as competition intensifies with Anthropic. At its Ignite conference, Microsoft presents new technology as a lower-cost alternative to leading offerings from OpenAI and Anthropic. OpenAI releases GPT-6.2 Astra for complex tasks and GPT-6.2 Sol for everyday use; both are available worldwide to paying ChatGPT and Codex subscribers, and OpenAI says they beat its earlier models on its benchmarks. The article also reports that Anthropic plans to become a public company, citing a reported $78 share price and a $1.4 trillion valuation target.

SID AI releases SID-2, an agentic search model that returns ranked document IDs, and SID-2 Swarm, which coordinates multiple agents. On the company’s 506-question benchmark across 19 collections, the 12-agent Swarm scores 0.720 NDCG; SID-2 Medium scores 0.585 in 3.11 seconds. The model has a 262,000-token context window and can target NDCG, F1 or Recall@k. SID says the results are self-reported, does not disclose model size or weights, and sets pricing from $0.25 per million input tokens. Separately, KnowMe Paper Radar ranks research papers using user preferences, recency, author and institutional context, and shows the sources behind its scores. Tests on 174,640 records found repeated context preparation was more than three times faster; cached abstracts lifted NDCG@10 from 0.412 to 0.598. It offers browser and command-line interfaces and does not use large-language-model inference in ordinary use.

S2W reports finding advertisements for AI-assisted exam cheating devices and proxy testing on Chinese-language Telegram channels, after reviewing dark-web, Telegram and community posts from November 16–17, 2026 with its DRI agent in the XARVIS platform. The largest channel it directly checked had more than 23,000 subscribers. A Chinese company claimed it helped the top scorer in South Korea’s certified tax accountant exam, but S2W could not verify that claim; it also warned that many sales posts appeared fraudulent, including promises to evade signal blockers and metal detectors.

In healthcare, Elevance Health says it aims to restore Medicaid profitability in 2027 by aligning reimbursement more closely with costs, tightening cost controls, and strengthening care management and behavioral-health oversight. It also says it is using responsibly governed AI to make healthcare navigation easier. In auto insurance, collision-industry leaders call for clearer rules as AI increasingly affects claims. Jill Tuggle cited an NAIC survey showing 88% of auto insurers use or are exploring AI, and said 25 states and Washington, D.C., adopted the NAIC model. She described a repair estimate of about $2,500 being reduced by AI to $438 before the claim was closed below the deductible. The model calls for risk-based controls, testing, monitoring, records, consumer notice and vendor oversight, but not claim-level notice, estimate-change logs or a named accountable reviewer. Tuggle says meaningful human review needs expertise, evidence, time and authority, including the ability to disagree without penalty.

Logistics companies report technology and AI returns but want vendors to demonstrate results sooner. In Sparq’s survey of 313 decision-makers, 61% of respondents whose returns met or exceeded expectations still wanted faster payback, while 87% expected budgets to grow next fiscal year. Legacy systems and integration were the leading barrier to faster investment (24%), followed by data quality (19%) and unclear returns (17%). Although 92% felt mostly or fully ready for the 2026 holiday shipping peak, 53% said forecasting accuracy was an area they were least prepared for; 34% expect dispatch and operations to see the greatest technology-driven change in two to three years. InnovateMR conducted the survey in August–September 2026, and return figures were self-reported, not independently audited.

PolyAI’s Dialog-RSN-1 is now available through AWS Marketplace and Amazon SageMaker JumpStart. The real-time conversational model processes caller audio and responds without a separate speech-to-text step. AWS customers can deploy it as an endpoint within their own account and connect through the OpenAI Realtime protocol; caller audio and transcripts stay within the customer’s AWS account, using its VPC, IAM roles and CloudWatch monitoring. Deployment is available through the SageMaker AI console or Python SDK, and the model is designed to respond to cues such as tone, hesitation and timing.

Key Takeaways

  • ServiceNow found AI-agent incidents at 92% of surveyed Indian organisations; only 36% contained an incident within 24 hours, and 67% lacked automated response systems.
  • Only 19% of Indian organisations named a person responsible for AI security, 18% had one oversight platform, and 37% discovered unregistered agents in production.
  • Nvidia reported $115.8 billion in quarterly revenue; its shares rose about 5% after hours. Its higher 2028 sales outlook and $150 billion buyback refocus attention on AI data-center investment.
  • Microsoft is pitching new Ignite technology as a lower-cost alternative to leading models from OpenAI and Anthropic; OpenAI launched GPT-6.2 Astra and Sol for paying ChatGPT and Codex subscribers.
  • SID AI’s 12-agent SID-2 Swarm scored 0.720 NDCG on its 506-question benchmark; SID-2 Medium scored 0.585 in 3.11 seconds. Pricing starts at $0.25 per million input tokens.
  • KnowMe Paper Radar tests on 174,640 records found repeated context preparation was over three times faster; cached abstracts raised NDCG@10 from 0.412 to 0.598.
  • S2W identified Telegram promotions for AI-assisted exam cheating and proxy testing, but could not verify a company’s claim that it helped South Korea’s certified tax accountant exam top scorer.
  • Elevance Health plans to restore Medicaid profitability in 2027 through closer cost-reimbursement alignment, cost controls and stronger care management, while also using governed AI for healthcare navigation.
  • Collision-industry leaders seek clearer AI claims rules; one example cited by Jill Tuggle involved a repair estimate of about $2,500 cut to $438 before a claim closed below the deductible.
  • In Sparq’s logistics survey, 87% expected technology and AI budgets to rise; legacy systems and integration were the top barrier to faster investment at 24%.
  • PolyAI’s Dialog-RSN-1 is available through Amazon SageMaker JumpStart and AWS Marketplace, processes caller audio without a separate speech-to-text step, and keeps audio and transcripts within the customer’s AWS account.

Indian Firms Struggle to Respond to AI Agent Security Incidents

A ServiceNow report found that 92% of organisations in India experienced at least one AI agent security incident in the past year. Only 36% resolved an incident within 24 hours. Just 19% assigned a dedicated person to AI agent security, and 18% had one platform to oversee all their AI agents. Although 51% monitored unusual activity in real time, 67% lacked automated response systems, while 37% discovered unknown AI agents on their systems.

ServiceNow Finds Most Indian Firms Cannot Quickly Contain AI Security Incidents

ServiceNow’s report, Deployed but Ungoverned: The State of Agentic Security in India, surveyed 500 security leaders. It found that 92% of Indian organisations experienced an AI agent security incident in the past year, but only 36% contained one within 24 hours. Only 19% had a person responsible for AI agent security, and 18% had a single platform to oversee their agents. The report said the gap between claimed security readiness and actual controls remains a concern.

ServiceNow Survey Finds AI Security Incidents at 92% of Indian Firms

A ServiceNow survey of 500 security leaders found that 92% of Indian organisations had an AI agent security incident in the past year, while only 36% contained one within 24 hours. The report, Deployed but Ungoverned: The State of Agentic Security in India, said only 19% had a named person responsible for AI security governance and 18% had one interface to see all their agents. Also, 37% discovered AI agents in production that security teams had not registered or known about. ServiceNow’s Kulmeet Bawa said organisations need to make agents visible, apply controls automatically and respond quickly.

Report Shows AI Agent Security Gaps at Indian Organisations

A ServiceNow study surveyed more than 3,700 senior security and technology decision makers in 20 countries, including 500 security leaders in India and 1,201 respondents across APAC. It found that 92% of Indian organisations experienced an AI agent security incident in the past year, but only 36% contained one within 24 hours. Half of Indian security leaders described their security as established, though the report said its foundations were uneven. ServiceNow’s Kulmeet Bawa said every AI agent should be registered, owned and visible in one place, with automatic controls and fast responses.

Nvidia’s Strong Results Put AI Infrastructure Stocks in Focus

Nvidia reported quarterly revenue of $115.8 billion, and its shares rose about 5% after hours. The article says leadership questions, investigations and export controls are changing expectations about where future AI data center spending may go. It looks at three stocks that may be affected by the AI infrastructure buildout and says a wider screen identified 51 more US and allied AI data center capacity builders.

Nvidia’s Spending Outlook Puts AI Data Center Stocks in Focus

Nvidia’s strong third-quarter results, higher 2028 sales outlook, and $150 billion buyback have drawn attention to spending on AI data centers. The article looks at three companies that could benefit from investment in the hardware and facilities behind AI. It says the same developments may encourage investors to buy or make them cautious. A broader screen found 49 other companies with AI data center infrastructure stories.

SID AI Launches SID-2 Search Model and 12-Agent Swarm

SID AI released SID-2, an agentic search model that searches data and returns ranked document IDs, and SID-2 Swarm, which uses multiple agents. SID reports that the model has a 262,000-token context window and can target NDCG, F1, or Recall@k scores. On its 506-question benchmark across 19 collections, the 12-agent Swarm scored 0.720 NDCG, while SID-2 Medium scored 0.585 in 3.11 seconds. The company says its results are self-reported, and the model’s size and weights are not disclosed. Pricing starts at $0.25 per million input tokens.

S2W Finds AI Exam Cheating Devices Sold on Telegram

South Korean data intelligence company S2W found posts advertising AI-assisted exam cheating devices and proxy testing on Chinese-language Telegram channels. It analyzed dark web, Telegram, and online community posts from November 16 to 17, 2026, using its DRI agent in the XARVIS platform. The largest channel S2W directly checked had more than 23,000 subscribers on November 17. A Chinese company claimed it had helped the top scorer in South Korea’s certified tax accountant exam, but S2W could not verify the claim. S2W said many sales posts appeared fraudulent, including claims that devices could evade signal blockers and metal detectors.

Elevance Health Plans to Rebuild Medicaid Profits by 2027

Elevance Health has outlined a plan to restore Medicaid profitability in 2027. The plan includes closer alignment of reimbursement with costs, tighter cost controls, and stronger care management and behavioral health oversight. The company is also using responsibly governed AI tools to make healthcare navigation easier. The article considers how these plans could affect its business and how investors view the company.

Microsoft and OpenAI Release New AI Models for Coding and Work

Microsoft and OpenAI released new AI models during a turbulent week for the industry. Microsoft introduced new technology at its Ignite developer conference and is presenting it as a lower-cost alternative to leading models from OpenAI and Anthropic. OpenAI released GPT-6.2 Astra for complex tasks and GPT-6.2 Sol as an everyday model. Both are available globally to paying ChatGPT and Codex subscribers, and OpenAI says they outperform its earlier models in its benchmarks. The article also notes Nvidia’s investor concerns and reports that Anthropic planned to become a public company, with a reported $78 share price and $1.4 trillion valuation target.

KnowMe Paper Radar helps researchers find promising papers

KnowMe Paper Radar is a lightweight tool that ranks research papers based on users’ topics, priorities, exclusions, and date ranges. It combines topic relevance, recency, author research histories, collaboration paths, and institutional context, while showing the sources behind each score. Tests on 174,640 records found that repeated context preparation was more than three times faster. Cached abstracts raised NDCG@10 from 0.412 to 0.598 on expert-reviewed retrieval tasks. A browser interface and command-line interface make the tool available to researchers and research agents, without using large language model inference in ordinary use.

Collision industry leaders call for clearer AI rules in insurance claims

At a Collision Industry Conference meeting in Las Vegas, Jill Tuggle of the Auto Body Association of Texas said AI now affects many parts of auto claims, but rules have not kept pace. She cited an NAIC survey showing that 88% of auto insurers use or are exploring AI, and said 25 states and Washington, D.C., adopted the NAIC model, which does not require a person to make the final decision on claim denials. Tuggle described unclear responsibility when AI changes repair estimates, including one claim where a shop’s estimate of about $2,500 was reduced to $438 and the claim was closed below the deductible. She called for consumers to be told when AI shapes a claim and who made the decision. The NAIC model calls for controls based on potential harm, model testing and monitoring, records, consumer notice, and vendor oversight, but does not require claim-level notice, an estimate change log, or a named accountable reviewer. Tuggle said meaningful human review should include competence, evidence, time, authority, and permission to disagree without penalty.

Logistics firms see AI returns but demand faster proof

Sparq’s first survey of 313 logistics decision-makers found that companies see returns from technology and AI but want vendors to prove results faster. Among respondents whose returns met or beat expectations, 61% also wanted faster return timelines, while 87% expected technology and AI budgets to rise next fiscal year. InnovateMR conducted the survey from August to September 2026, and its return figures were self-reported rather than independently audited. Respondents named legacy systems and integration as the top barrier to faster AI investment at 24%, followed by data quality at 19% and unclear returns at 17%. Although 92% felt mostly or fully prepared for the 2026 holiday shipping peak, 53% named forecasting accuracy as a challenge they were least prepared for. Looking two to three years ahead, 34% expected dispatch and operations to see the greatest technology-driven change.

PolyAI brings its Dialog-RSN-1 model to Amazon SageMaker JumpStart

PolyAI’s Dialog-RSN-1 reasoning model for real-time conversations is now available through AWS Marketplace and Amazon SageMaker JumpStart. The model processes a caller’s audio, reasons about the caller’s needs, and responds in real time without a separate speech-to-text step. AWS customers can deploy it as an endpoint inside their own account and connect it through the OpenAI Realtime protocol. The endpoint uses the customer’s virtual private cloud, IAM roles, and CloudWatch monitoring, keeping caller audio and transcripts within the customer’s AWS account. Users can deploy it through the SageMaker AI console or with the SageMaker Python SDK. The model is designed to recognize tone, hesitation, and timing while responding quickly during live calls.

Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

AI agent security AI governance ServiceNow AI incident response Nvidia AI data centers Microsoft OpenAI Anthropic GPT-6.2 agentic search SID-2 Swarm research paper ranking AI-assisted exam cheating healthcare AI AI insurance claims logistics technology conversational AI PolyAI AWS SageMaker

Comments

Loading...