According to OpenAI, enterprise AI has graduated from the sandbox and is now used in daily operations with tight workflow integration.

New data from the company shows that companies are now assigning complex, multi-step workflows to models, rather than simply requiring a text summary. These diagrams illustrate a major shift in the way organizations are implementing generative models.

OpenAI’s platform currently serves more than 800 million users each week, with a “flywheel” effect that familiarizes consumers with professional environments. The company’s latest report states that more than 1 million enterprise customers are currently using these tools, and the goal is even tighter integration going forward.

This evolution presents two realities to decision makers. Although productivity gains are tangible, the widening gap between “frontier” adopters and median firms suggests that value is highly dependent on intensity of use.

From chatbots to deep reasoning

The best metric for determining a company’s implementation maturity is task complexity, not seat count

OpenAI reports an 8x year-over-year increase in ChatGPT message volume, but a better indicator for enterprise architects is the consumption of API inference tokens, which suggests deeper integration is taking place. This number has increased nearly 320x per organization. This is evidence that companies are systematically building more intelligent models into their products to handle logic rather than basic queries.

The rise of configurable interfaces supports this view. Weekly users of Custom GPT and Custom Projects (tools that allow employees to direct models with specific organizational knowledge) have increased approximately 19x this year. Approximately 20% of all corporate messages are now processed through these customized environments, making standardization a prerequisite for professional use.

For company leaders auditing the ROI of AI seats, the data provides evidence of time savings. On average, users believe this technology saves them 40 to 60 minutes per active day. Effects vary depending on the type of job. Data science, engineering, and communications professionals report even greater savings (60-80 minutes per day on average).

Beyond efficiency, software is changing the boundaries of roles. There are certain implications for technical capabilities, especially regarding code generation.

OpenAI says enterprise users are seeing an increase in coding-related messages across all business functions. Coding queries for non-engineering, IT, and research roles have increased by an average of 36% over the past six months. Non-technical teams are using this tool to perform analysis that previously required professional developers.

Business improvements spread throughout the department. Survey data shows that 87% of IT professionals report faster problem resolution and 75% of HR professionals see an increase in employee engagement.

The widening enterprise AI capability gap

OpenAI’s data suggests a divide is emerging between organizations that simply provide access to tools and those where integration is deeply embedded in their operating models. The report identifies “frontier” workers, or workers at the 95th percentile of adoption intensity, who generate six times more messages than the median worker.

This disparity is evident at the organizational level. Frontier companies generate approximately twice as many messages per sheet as the median company and seven times as many messages for custom GPT. Larger companies aren’t just using tools more frequently; They are investing in the infrastructure and standardization needed to make AI a permanent part of their operations.

Users who tackle a variety of tasks (approximately seven types) report saving five times more time than those who limit their usage to three or four basic functions. Benefits are directly correlated to intensity of usage, suggesting that a “light touch” implementation plan may not achieve the expected ROI.

Professional services, finance, and technology sectors were early adopters and remain the biggest adopters, but other industries are doing their best to catch up. The technology sector leads with 11x year-over-year growth, followed by healthcare and manufacturing with 8x and 7x growth, respectively.

The global adoption pattern also challenges the notion that this is simply a US-centric phenomenon. International usage is rapidly increasing, with business customer growth exceeding 140% year over year in markets such as Australia, Brazil, the Netherlands, and France. Japan has also emerged as an important market with the largest number of corporate API customers outside the US.

OpenAI: Tight AI integration accelerates enterprise workflows

Case studies highlight how these tools impact key business metrics. Retailer Lowe’s implemented an employee tool in more than 1,700 stores, resulting in a 200 basis point improvement in customer satisfaction scores when employees used the system. Additionally, when online customers utilized the retailer’s AI tools, conversion rates more than doubled.

In the pharmaceutical space, Moderna used enterprise AI to speed up the creation of target product profiles (TPPs), a process that typically takes weeks of cross-functional effort. The company shortened core analysis steps from weeks to hours by automating the extraction of key facts from large packs of evidence.

Financial services company BBVA leveraged this technology to fix bottlenecks in corporate signature authority legal verification. By building a generative AI solution to handle standard legal queries, the bank automated over 9,000 queries per year, effectively freeing up the equivalent of three full-time employees for higher-value tasks.

However, moving to production-grade AI requires more than just software procurement. It requires organizational preparation. The main hurdle for many organizations is no longer the functionality of the model, but its implementation and internal structure.

Leading companies consistently enable tight system integration by turning on connectors that provide models with secure access to corporate data. However, approximately one in four companies do not take this step, limiting their models to general knowledge rather than specific organizational contexts.

Successful implementation requires executive support to set explicit mandates and encourage the codification of organizational knowledge into reusable assets.

As technology continues to evolve, organizations must adjust their approach. According to data from OpenAI, success now depends on delegating complex workflows with deep integrations and treating AI as the primary engine of revenue growth for enterprises, rather than just looking for output.

See also: AWS re:Invent 2025: Frontier AI agents replace chatbots

Want to learn more about AI and big data from industry leaders? Check out the AI ​​& Big Data Expo in Amsterdam, California, and London. This comprehensive event is part of TechEx and co-located with other major technology events including Cyber ​​Security Expo. Click here for more information.

AI News is brought to you by TechForge Media. Learn about other upcoming enterprise technology events and webinars.

Resemble AI has raised $13 million in a new strategic investment round for AI deepfake detection. This funding brings the company’s total venture investment to $25 million, with participation from Berkeley CalFund, Berkeley Frontier Fund, Comcast Ventures, Craft Ventures, Gentree, Google’s AI Futures Fund, IAG Capital Partners, and others.

The funding comes as organizations are under pressure to verify the authenticity of digital content. Generative AI has made it easier for criminals to create convincing deepfakes, resulting in over USD 1.56 billion in fraudulent losses in 2025. Analysts estimate that generative AI could lead to US$40 billion in fraud losses in the US by 2027.

Recent incidents highlight how quickly threats can evolve. In Singapore, 13 people collectively lost more than S$360,000 after fraudsters impersonated telecommunications providers and the Monetary Authority of Singapore. Attackers used caller ID spoofing, audio deepfakes, and social engineering techniques to create urgency and capitalize on the public’s trust in governments and telecom brands.

Deepfake detection tools and new AI features

Resemble AI develops real-time verification tools that enable enterprises to detect AI-generated audio, video, images, and text. The company plans to use the new funding to expand global access to its AI deepfake detection platform. This includes two recent releases:

• DETECT-3B Omnia deepfake detection model designed for enterprise environments. The company reports 98% detection accuracy in over 38 languages.
• similar to intelligenceis a platform that uses Google’s Gemini 3 model to provide explainability for multimodal AI-generated content.

Resemble AI positions these tools as part of a broader effort to support real-time validation of human users and AI agents interacting with digital content.

The company says the DETECT-3B Omni is already being used in areas such as entertainment, communications and government. Public benchmark results for Hugging Face show that this model performs best in image and audio deepfake detection, with a lower average error rate than competing models.

Industry participants say rapid improvements in generative AI are changing the way companies think about content trust and identity systems. Representatives from Google’s AI Futures Fund, Sony Ventures, and Okta noted that organizations are moving to a verification layer that helps maintain trust in the authentication process.

Alongside the investment announcement, Resemble AI released its outlook on how deepfake-related risks will evolve in 2026. The company anticipates several changes that could shape its corporate plans.

Deepfake verification could become standard for official communications

The company predicts that real-time deepfake detection may eventually be required in official video conferences following incidents involving government officials. Such a move could create new procurement activities and increase adoption in the public sector.

Organizational readiness can determine competitive position

As more jurisdictions adopt AI regulations, companies that integrate training, governance, and compliance processes early may be better able to respond to operational and regulatory demands.

Identity emerges as a central focus for AI security

As many AI-related attacks rely on impersonation, organizations are likely to place greater emphasis on identity-centric security models, such as zero-trust approaches for human and machine identities.

Cyber ​​insurance costs may rise

The rise in corporate deepfake incidents may cause insurance companies to reconsider the insurance policies they offer. Businesses without detection tools may face higher premiums and coverage limitations.

This investment highlights the growing need for companies to understand how generative AI changes their risk exposure. Organizations across all sectors are evaluating how verification, identity protection, and incident response can be incorporated into broader security and compliance strategies.

(Photo courtesy of Pau Casals)

Reference: AWS re:Invent 2025: Frontier AI agents replace chatbots

Want to learn more about AI and big data from industry leaders? Check out the AI ​​& Big Data Expo in Amsterdam, California, and London. This comprehensive event is part of TechEx and co-located with other major technology events. Click here for more information.

AI News is brought to you by TechForge Media. Learn about other upcoming enterprise technology events and webinars.

Digital payments and fintech company Ant International has won the NeurIPS competition for fairness in AI facial detection. The company says it is focused on developing secure and inclusive financial services, especially as deepfake technology becomes more commonplace.

The increasing use of facial recognition in many fields has highlighted the issue of algorithmic bias in AI. A study conducted by NIST (National Institute of Standards and Technology) found that many widely used facial recognition algorithms have significantly higher error rates when analyzing the faces of women and people of color, and this disparity is due to a lack of diversity in the training data and demographics of the people building and controlling many mainstream AI platforms. The results of biased algorithms can lead to the denial of financial services to large sections of the population and are considered weaknesses in security protocols.

The NeurIPS competition was held in parallel with the acclaimed AI conference, the Conference on Neural Information Processing Systems, and challenged participants to create high-performance and fair AI models that cover a variety of demographic factors, including gender, age, and skin color. The Ant International team beat out more than 2,100 entries from 162 teams from around the world. The task given was to accurately detect 1.2 million AI-generated facial images that were selected to be reasonably representative of demographic groups.

The approach taken by Ant’s superior AI models combines a Mixture of Experts (MoE) architecture with bias detection mechanisms. This system trains two competing neural networks. One focuses on identifying deepfakes, while the other is designed to counter the former, forcing them to ignore demographic characteristics. This dynamic process ensures that the system learns how to detect true signs of manipulation, rather than carelessly relying on demographic patterns. The model was trained on globally representative datasets and incorporated real-world payment fraud scenarios to ensure performance at scale.

“Biased AI systems are inherently insecure,” explains Dr. Tianyi Zhang, general manager of risk management and cybersecurity at Ant International. “The fairness of our model is not just an ethical issue; it is fundamental to preventing deepfakes from being exploited and ensuring reliable identity verification for all users.”

The technology behind the winning entry is now being integrated into Ant’s payments and financial services to combat the threat of deepfakes, and the companies say it has achieved a detection rate of over 99.8% across all demographics and the 200 markets in which Ant operates.

Ant’s technology helps customers meet global electronic know-your-customer (eKYC) standards without algorithmic bias, especially during customer onboarding. This is considered to be particularly important in emerging markets where greater financial inclusion may be hindered.

Ant International serves more than 150 million merchants and 1.8 billion user accounts, known for services such as Alipay+, Antom, Bettr, and WorldFirst. The company says AI security is a pillar of its business. AI SHIELD is a risk management framework built on AI Security Docker that reduces the risk of AI service vulnerabilities such as unauthorized access and data leakage.

AI SHIELD powers a suite of risk management solutions that provide broad protection for financial transactions, including protection against deepfake attacks and fraud. The company says that Alipay+ EasySafePay 360 has reduced account takeover incidents in digital wallet payments by 90%.

(Image source: “Beautiful Portrait Abstract Art, Three-dimensional Shapes, Geometric Shapes, Neo-Tokyo Colors, Subdued Colors, Pixar, Art Station, Greg Rutkowski, Samdosart, GE” – Public Domain)

Want to learn more about AI and big data from industry leaders? Check out the AI ​​& Big Data Expo in Amsterdam, California, and London. This comprehensive event is part of TechEx and co-located with other major technology events. Click here for more information.

AI News is brought to you by TechForge Media. Learn about other upcoming enterprise technology events and webinars.

When ByteDance launched a prototype agent-based AI smartphone with ZTE on December 2nd, it quickly sparked consumer enthusiasm, but just as quickly, privacy concerns arose, forcing the company to dial back capabilities. But behind the headline-grabbing sell-out and ensuing controversy lies a more important story. That is the enterprise impact of operating system-level AI agents that can autonomously perform complex multi-step tasks in device ecosystems.

Powered by ByteDance’s Doubao large language model, the ZTE Nubia M153 represents both an experiment in a consumer gadget and a preview of how agentic AI smartphones can reshape workplace productivity, field operations, and enterprise mobility strategies, if the technology can overcome the fundamental trust and governance challenges required for enterprise adoption.

From consumer curiosity to corporate necessity

The appeal for consumers is clear: restaurant reservations by voice, automatic photo editing, cross-platform price comparisons, and more. However, Gartner predicts that by 2028, 33% of enterprise software applications will include agent AI capabilities, up from less than 1% in 2024.

As the most ubiquitous computing device in enterprise workflows, smartphones provide an important testing ground. “Agent AI in industries like manufacturing, construction, healthcare, and energy enhances decision-making, improves safety, and streamlines tasks,” said Nicholas Muy, CISO at Scrut Automation. However, he cautions that early adopters need to avoid real risks around AI errors and security gaps.

According to McKinsey research, 23% of organizations are extending agent AI systems in at least one business function, and an additional 39% are experimenting with AI agents. However, enterprise deployment is different from consumer use. Governance frameworks, audit trails, role-based permissions, and compliance mechanisms are needed, and ByteDance’s consumer prototype was notably lacking.

China’s strategic advantage in software and hardware integration

ByteDance’s approach (partnering with ZTE rather than building its own hardware) reflects a successful enterprise AI strategy. The company is positioning Doubao as a system-level integration that any manufacturer can adopt, similar to what Google did with Android.

With 157 million monthly active users as of August 2025, Doubao already dominates China’s consumer AI market, more than double Tencent’s Yuanbao, which has 73 million users, according to QuestMobile data.

The software over hardware strategy addresses what Morgan Stanley analysts have identified as key weaknesses. The idea is that major smartphone makers, including Apple, Huawei and Xiaomi, have strong enough technological capabilities to develop AI assistants in-house rather than partnering with third-party providers.

ByteDance’s realistic target market appears to be secondary manufacturers seeking differentiated capabilities and potentially enterprise device management platforms. For business buyers, this fragmentation presents both opportunities and challenges.

Organizations can choose device manufacturers based on hardware requirements while standardizing AI capabilities, but only if the governance and security framework proves to be robust enough for the regulated industry.

Privacy panic reveals corporate requirements

Rapid backlash after entrepreneur Taylor Ogan demonstrated the M153’s capabilities on social media and revealed what companies are looking for in hiring it. When users witnessed AI agents with deep system privileges autonomously accessing apps, processing payments, and manipulating data, their immediate concern was control, not convenience.

Trust remains the main barrier to adoption, according to a Forum Ventures survey of 100 senior enterprise IT decision makers. “The trust gap is huge,” says Jonah Midanik, general partner at Forum Ventures. “AI agents can perform tasks with incredible efficiency, but their output is based on statistical probabilities rather than inherent truths.”

The feature rollbacks reported by ByteDance demonstrate an understanding that enterprise-grade agent-based AI smartphones require fine-grained permission systems, comprehensive logging, and the ability to define strict operational boundaries, features that are notably lacking in consumer prototypes.

Enterprise and consumer: different use cases, different requirements

Agent AI The enterprise use case for smartphones is very different from consumer applications. Field service technicians can use AI agents to proactively uncover equipment history, recommend the best route based on real-time conditions, and guide them through complex steps without manual searches. Healthcare providers can access patient context and decision support without having to navigate multiple systems. Financial services professionals receive compliance-checked recommendations and automated workflow orchestration.

According to PwC research, 79% of organizations have deployed AI agents at some level, and 96% of IT leaders plan to expand in 2025. However, a Cloudera survey of 1,484 IT decision makers found that successful enterprise adoption requires a gradual rollout with industry-specific data integration, transparent decision-making processes, and comprehensive testing.

The consumer smartphone market, which IDC predicts will ship 912 million generated AI-enabled units by 2028, is focused on personalization and convenience. Auditability, compliance, and risk mitigation are priorities for enterprise deployments, requirements that are not yet met by consumer-grade agent-based AI smartphones.

Global competitive dynamics and regional strategies

The technology gap between the United States and China adds further complexity. Apple’s slow rollout of Apple Intelligence in mainland China has left ByteDance, Alibaba, Baidu and Tencent racing to fill the void. But Apple’s approach is fundamentally different. Tight hardware and software integration with on-device processing prioritizes user privacy, a stance that resonates with enterprise security requirements.

ByteDance’s licensing strategy could allow Doubao to quickly penetrate the market with Chinese manufacturers and establish the de facto standard before Western competitors can match operating system level integration. For multinational companies operating in the region, this creates device management challenges around data sovereignty, compliance frameworks, and a consistent user experience.

According to Counterpoint Research, Asia Pacific is the fastest growing market for AI agents, with the US currently accounting for 40.1% revenue share. Enterprise buyers may maintain separate device strategies for different regulatory environments and must navigate a bifurcated landscape.

The way forward: Solutions to overcome the hype

For enterprise leaders evaluating agent AI smartphones, ByteDance’s prototype offers valuable lessons on what to ask of vendors.

First, it’s a comprehensive governance framework that defines decision boundaries, records all autonomous actions, and provides role-based access controls. Anthropic’s enterprise solutions feature centralized provisioning, audit logging, and role-based entitlement and may demonstrate these requirements.

The second is a hybrid approach that balances on-device processing for sensitive operations with cloud capabilities for complex inference. Enterprise deployments require flexibility to meet varying data residency and compliance requirements within a jurisdiction.

Third, roll out gradually starting with low-risk use cases. Amazon’s introduction of AI agents to modernize Java applications shows how enterprises can capture value while managing risk.

ByteDance and ZTE’s partnership ultimately heralds an inevitable convergence. In other words, agent AI capabilities will become a standard feature on smartphones rather than a premium differentiator. Enterprise adoption follows a proven pattern of pilot programs in controlled environments, security validation, and gradual expansion as the governance framework matures.

The question facing enterprise technology leaders is not whether agent-based AI smartphones will impact productivity in the workplace, but rather whether to proactively develop an adoption strategy or whether enterprise capabilities will react to improved consumer technology. The privacy panic that followed ByteDance’s launch suggests that organizations that demand enterprise-grade security and governance from the start will determine the trajectory of this technology.

Gartner predicts that by 2028, at least 15% of work decisions will be made autonomously by agent AI, but none by 2024, making the smartphone a communication device and an autonomous enterprise agent. The winners will not be those who adopt first, but those with the most thoughtful implementations that incorporate security and scalable governance from day one.

SEE ALSO: IBM cites agent AI, data policy and quantum as trends for 2026

Want to learn more about AI and big data from industry leaders? Check out the AI ​​& Big Data Expo in Amsterdam, California, and London. This comprehensive event is part of TechEx and co-located with other major technology events. Click here for more information.

AI News is brought to you by TechForge Media. Learn about other upcoming enterprise technology events and webinars.