AI Insights: Key Global Developments in February 2026

Welcome to the February 2026 edition of our global AI update.

This month feels different. The headlines are still about bigger models and new releases, but the real story is what’s happening behind the scenes. Companies are building serious infrastructure. Enterprises are putting AI directly into planning, marketing, data, and customer systems. Governments are moving from discussions to clear rules.

It’s less about experimentation now and more about execution. AI is slowly becoming part of everyday business systems, not just a tool teams test on the side.

Here are the major moves worth noting:

OpenAI - GPT-5.3 Codex (Agentic Code Model)

OpenAI announced GPT-5.3-Codex, a major upgrade to its coding models. This new model combines the coding prowess of GPT-5.2-Codex with the deep reasoning of GPT-5.2, and runs ~25% faster. It achieves state-of-the-art results on industry coding benchmarks (SWE-Bench Pro, Terminal-Bench) and can autonomously execute long-running development tasks. 

Notably, GPT-5.3-Codex is described as the first model “instrumental in creating itself,” since OpenAI’s team used early versions of Codex to debug and manage its own training. In practice, GPT-5.3-Codex can plan, write, and iterate on complex software projects (the blog shows it building games with minimal prompting).

This marks a new high-water mark for AI coding assistants and agentic developer tools.

Source - OpenAI

Meta Platforms - Superintelligence Lab Models

At the Davos Forum, Meta’s CTO revealed that Meta’s new “Superintelligence” research lab has internally built its first high-profile AI models. Although not officially released yet, these prototypes (codenamed Avocado for text and Mango for image/video) reportedly outperform earlier Llama models on large-scale benchmarks.

Meta says the early versions are “very good” and on par with the latest models from other AI leaders. 

This suggests Meta is closing the gap in core model capabilities, even as it wrestles with delivering them responsibly. We expect more details on Meta’s next-gen models later this year, but these announcements show major labs still racing to define the cutting edge.

Source- CXO

Chinese AI Models - GLM-5, Seedance 2.0, Kling 3.0

In China, startups and tech giants also announced new AI models. On Feb 11, Zhipu AI (known for the GLM line) released GLM-5, a 128B-parameter open-source model tailored for coding and agentic tasks.

Zhipu claims GLM-5 surpasses Anthropic’s Claude Opus 4.5 on standard benchmarks, and was trained entirely on domestic chips (Huawei’s Ascend, Moore Threads, Cambricon, Kunlunxin) due to export controls. Meanwhile ByteDance and Kuaishou unveiled vision models: ByteDance’s Seedance 2.0 (video generation) and Kuaishou’s Kling 3.0 (multimodal). 

These launches underscore China’s push to develop its own large-model stack and compete in generative AI, especially for video and creative content. Together, these Chinese efforts highlight both technological progress and the strategic drive for self-reliance in AI infrastructure.

Source- Zhipu

Board & Microsoft - Agentic Planning Agents

Enterprise-planning vendor Board introduced “Board Agents”, AI assistants for corporate planning built on Microsoft’s Foundry platform.

The initial release includes AI agents for financial planning & analysis (FP&A) and controlling, with more for merchandising and supply chain to follow. These agents run on Azure and are natively embedded in Board’s planning software, so they use the company’s own data and models. For example, the FP&A Agent can take a balance sheet and income statement and autonomously generate actionable insights. 

This move illustrates how enterprises are embedding AI into core workflows: the agents are not standalone chatbots but fully integrated planning tools, complete with governance and audit trails.

The Board says the goal is to deliver faster ROI by having AI “understand customers’ data, processes, and policies” rather than bolt-on generative assistants.

Source- Microsoft

Franklin Templeton & Microsoft - AI-Powered Distribution Hub

Asset manager Franklin Templeton unveiled “Intelligence Hub”, a wealth-distribution platform powered by Microsoft Azure. This modular system centralizes sales and client data (from CRM, research, market feeds) into one interface and adds AI-driven workflows. 

Built on the Azure stack (including Foundry and Dynamics 365), the Hub automates tasks like lead list generation and meeting prep, freeing advisors to focus on high-value client interactions. In pilots, Franklin Templeton saw measurable gains: reduced prep time and more qualified client engagements. The launch is part of a multiyear strategy (announced in 2024) to apply responsible AI across their enterprise. 

According to the company, Intelligence Hub “raises a new benchmark” by using AI to personalize experiences at scale, reflecting the broader trend of embedding AI into business workflows rather than just offering standalone tools.

Source- Franklin Templeton

NTT DATA & AWS - Enterprise AI Cloud Collaboration

NTT DATA announced a multi-year Strategic Collaboration Agreement with Amazon Web Services. Under this deal, NTT DATA will help enterprise customers modernize legacy systems on AWS and “adopt agentic AI responsibly” with cloud solutions. 

The partnership targets four areas: large-scale, AI-driven cloud migrations; industry-specific cloud offerings (financial services, healthcare, etc.); managed AI and data services; and a European sovereign cloud for regulated workloads.

For example, NTT DATA will build pre-packaged “industry cloud” templates with embedded AI agents, and will extend AWS’s contact-center solutions with generative AI for customer experience. The alliance also commits thousands of AWS-certified experts to accelerate AI adoption. 

In short, this arrangement signals how system integrators are positioning themselves: bridging AI and cloud at scale for enterprises, especially in regulated sectors

Source- NTT DATA

Snowflake & OpenAI- AI Meets Data Platforms

Snowflake and OpenAI announced a $200 million partnership to integrate OpenAI models directly into Snowflake’s data platform. Under this agreement, OpenAI’s API becomes a built-in capability of Snowflake’s AI services (Snowflake Cortex AI and Snowflake Intelligence). 

This means Snowflake customers can query their own data with GPT-5.2 via natural language and deploy custom AI agents grounded in enterprise data.

For example, analysts could ask business questions in English (“Show me our latest sales trends”), and the system would automatically retrieve and analyze the data without writing SQL. This tight integration brings “frontier intelligence” to enterprise data: companies like Canva and WHOOP are already testing it to automate insight generation and support daily decisions. 

In essence, Snowflake is making it easy for its 12,600+ customers to use OpenAI’s models within familiar data workflows, reflecting the growing norm of blending LLMs with large-scale data platforms.

Source- OpenAI

Cognizant & Typeface - AI for Marketing

IT services firm Cognizant and marketing AI startup Typeface announced a strategic partnership to modernize marketing operations. Typeface provides an “agentic AI orchestration” platform for marketing workflows. 

Under the deal, enterprises can use Typeface to automate the full marketing lifecycle – from content ideation and creation to multi-channel optimization- with AI agents coordinating the process. Cognizant will supply consulting, creative and implementation services to help clients adopt these AI-driven workflows.

In practice, this means companies can connect Typeface’s platform with existing CRM and CMS systems, then spin up AI agents that generate campaign ideas, produce copy, and adjust distribution in real time. 

The partnership highlights how firms are commercializing AI not just for IT or CX, but for business functions like marketing; it promises a more “modular, software-driven” marketing operation at scale. 

Source- Cognizant

Medallia & Ada- AI-Driven Customer Experience Automation

Medallia, a customer experience platform and AI chatbot company Ada announced a joint solution for contact centers.

The idea is to connect Medallia’s analytics (surveys, call transcriptions, social feedback) with Ada’s conversational AI agents. Medallia will ingest Ada’s chat transcripts and other signals into its analytics engine, then feed prioritized insights into Ada’s AI-powered automation. 

In practice, this lets the AI agents take automated actions (like resolving issues or guiding purchases) based on real-time customer feedback and risk scores. For example, if Medallia flags a surge in complaints about a particular product, Ada’s agent could proactively reach out to affected customers via chat. 

According to the companies, this “insights-to-action” integration means CX leaders can move beyond isolated pilots: the joint system closes the loop between listening to customers and acting on that insight automatically.

Source- Medallia

NVIDIA & CoreWeave- AI Data Center Expansion

NVIDIA and cloud provider CoreWeave announced an expanded collaboration to build out massive AI data centers. Nvidia is investing $2 billion in CoreWeave, which specializes in “AI factories” (data center campuses) across the US. 

As part of the deal, CoreWeave will deploy multiple generations of Nvidia hardware (including upcoming Rubin and BlueField GPUs) in new AI-optimized facilities totaling over 5 GW by 2030.

The partnership even includes a $450M equity stake in CoreWeave by Nvidia. This is essentially Nvidia locking in long-term demand for its chips and advancing scalable AI infrastructure: Nvidia’s roadmap will now be co-optimized with a major cloud customer. It signals that tech leaders see on-prem AI compute capacity and dedicated data-center buildout (often powered by renewable energy) as strategic priorities.

Source- NVIDIA

Meta - $10B Indiana AI Data Center

Meta (Facebook’s parent) broke ground on a $10 billion, 1 GW AI data center in Lebanon, Indiana. This massive facility (expected online in 2027-28) will support Meta’s AI ambitions by providing huge computing capacity locally. 

The project is part of Meta’s recent $600B pledge for U.S. infrastructure (mostly data centers and servers). Like Microsoft, Google and others, Meta is racing to build an exascale compute for generative AI-  a “once-in-a-generation” expansion. This deal underscores how hyperscalers are securing on-site power and land for specialized AI chips (Meta even has $27B lined up for another 2 GW site in Louisiana). 

For enterprises, the takeaway is clear: the cloud isn’t the whole story- owning dedicated, AI-tuned infrastructure (even under companies’ own roofs) is becoming just as crucial for leading-edge AI work.

Source- Meta

European Union - Data Omnibus & AI Code Taskforce

In EU policy, regulators are threading the needle between innovation and oversight. The EU data authorities (EDPB/EDPS) published a joint opinion on the Commission’s “Digital Omnibus” proposals.

They broadly support aligning AI rules with existing laws (e.g. raising breach thresholds, clarifying data use for R&D) but warn against weakening core privacy protections (for instance, by too narrowly defining “personal data”). These opinions advise the EU to ease compliance burdens (for example, longer timelines for high-risk AI rules) without undermining user rights.

Separately, the EU’s new AI Office convened the first meeting of signatories to the General-Purpose AI Code of Practice on Jan 30. This taskforce of companies and experts will help implement the voluntary code (covering principles like transparency and safety) as a complement to the binding AI Act.

The existence of these codes and taskforces means firms should stay engaged: even as the formal AI Act may see delayed deadlines, companies are being nudged to adopt best practices early.

Source- EDPB

United States- Federal AI Executive Order

The US is still in the early stages of federal AI policy. Last December’s Executive Order (EO 2025-12) established a national AI strategy, including an “AI Litigation Task Force” to challenge conflicting state laws. 

By Feb, agencies were implementing this roadmap: for example, the EO mandates a review of state AI rules (to be completed by March 2026) and preempts overly restrictive state legislation (e.g. California’s proposed AI hiring law).

In practice, we expect the White House and DOJ will start scrutinizing state AI regulations. For US enterprises, the key is to follow federal guidance (NIST’s AI security standards, OMB memos, etc.) and be mindful that state rules might change or face preemption.

Source- The White House

India- AI Content Rules and Summit

India moved on two fronts. 

First, on Feb 10 the government published new IT Rules explicitly targeting AI-generated content. The rules (effective Feb 20) require platforms to label AI audio/video and remove illegal content within 3 hours (2 hours for intimate media).

This is a sharp tightening: non-compliance can now bring heavy penalties under India’s IT Act. Companies operating in India should update their content-moderation systems accordingly and build in provenance metadata for AI outputs.

Second, India is hosting the first Global AI Impact Summit (Feb 16-20 in New Delhi). This event (the first major summit in the Global South) will convene leaders from ~100 countries around themes of human-centric AI, sustainability, and inclusion.

The outcome is likely to emphasize collaborative principles (e.g. equitable AI access, ethical norms) that could influence future regulations. Multinational firms should be aware that India is positioning itself as a rule-making voice, so engagement (and alignment with “AI for All” principles) will be important.

Source- MeitY, IndiaAI

To Wrap Things Up

February made one thing clear that AI is settling into real-world operations. The companies that stand out right now are not just launching new models, they’re building data centers, forming long-term partnerships, and connecting AI directly to their core workflows. And at the same time, regulators are setting clearer expectations, which means compliance can’t be an afterthought anymore.

As we move forward, the gap will grow between organizations that treat AI as a side feature and those that redesign how work gets done around it. The second group will move faster and scale with more confidence.

Stay informed with our regulatory updates and join us next month for the latest developments in risk management and compliance!

For any feedback or requests for coverage in future issues (e.g., additional countries or topics), please contact us at info@riskinfo.ai. We hope you found this newsletter insightful.

Best regards,

The RiskInfo.ai Team