It’s a wrap: News this week (Aug 17-21)

Enterprise technology developments this week reflected the growing push to move AI into production, spanning business workflows, drug development, connected vehicles and data engineering. At the same time, security incidents and acquisitions highlighted the new risks created by increasingly autonomous AI systems. Here are some of the top enterprise technology news in India and across the world.

Hexaware responds to data-leak claim, launches Zero Vulnerability

Hexaware Technologies said its internal investigation found “no reliable evidence” of a breach after a threat actor allegedly claimed exposure of certain employee information. In an August 19 regulatory filing, the IT services company said it was responding to reports about the alleged incident but did not disclose the nature or volume of the information involved.

Separately, Hexaware launched Zero Vulnerability, a cybersecurity offering designed to help enterprises identify, prioritise and remediate vulnerabilities. The company said AI-based security tools are increasing the speed at which vulnerabilities are discovered, but enterprise security and engineering teams still have limited capacity to investigate and fix them.

The offering combines continuous discovery, AI-assisted prioritisation, automated remediation and engineering support. It is aimed at reducing the backlog between vulnerabilities detected and those resolved as enterprises expand their cloud, application and AI estates. Hexaware’s filing

Tech Mahindra, ServiceNow expand partnership to scale enterprise AI

Tech Mahindra and ServiceNow have expanded their multi-year partnership to help enterprises move artificial intelligence projects from pilots to production-scale deployments. The companies will combine Tech Mahindra’s consulting, engineering and implementation capabilities with the ServiceNow AI Platform to develop industry-specific solutions and enterprise automation workflows.

The partnership will focus on manufacturing, telecommunications, banking and financial services, media and technology. Tech Mahindra also plans to expand its ServiceNow practice and establish an AI and innovation centre of excellence.

A key component is the “Client Zero” approach, under which Tech Mahindra and the Mahindra Group will act as large-scale validation environments for ServiceNow-based AI deployments. The companies intend to use lessons from these implementations to develop reusable playbooks for customers, covering adoption, governance, integration and measurable business outcomes. The partnership reflects growing enterprise demand for production-ready AI rather than standalone proofs of concept. (Read more)

TCS launches agentic AI platform for drug development

Tata Consultancy Services has launched ADD AgentHub, a role-based agentic AI platform designed for pharmaceutical research, clinical trials and pharmacovigilance. The platform enables pharmaceutical companies to configure and deploy AI agents across clinical-development and drug-safety workflows while retaining human oversight and auditability.

The agents can support activities including clinical data review, study design, protocol digitisation, medical monitoring and safety-case processing. TCS said the platform could deliver efficiency improvements of up to 40% in clinical data management, reduce study-build effort by 30% and lower safety-case processing costs by as much as 30%.

ADD AgentHub follows a “human plus AI” operating model in which agents perform defined operational tasks while human specialists remain accountable for decisions. The governance structure is intended to address regulatory and compliance requirements as pharmaceutical companies scale AI across sensitive research and patient-related processes. TCS announcement. (Read more)

AWS brings OpenAI’s GPT-5.6 models to India with local inference

Amazon Web Services has made OpenAI’s GPT-5.6 Terra and Luna models available through Amazon Bedrock in India, enabling enterprises to process AI workloads locally across its Mumbai and Hyderabad regions. The move targets banks, healthcare providers, public-sector organisations and other businesses with data-residency and compliance requirements. Terra is designed for general-purpose enterprise applications, while Luna supports high-volume, latency-sensitive tasks such as summarisation and classification. AWS said the launch also reflects recent price reductions of up to 20% for Terra and 80% for Luna. The availability could help Indian enterprises move sensitive AI applications from experimentation into production.

Google Cloud, Mahindra bring Gemini into connected vehicles

Mahindra has partnered with Google Cloud to introduce a conversational automotive assistant in its BE 6 SPORTEQ electric vehicle. The assistant is built using Google Cloud’s Gemini Enterprise models and Mahindra’s artificial intelligence architecture, MAIA.

Unlike conventional in-car voice systems that depend on predefined commands, the assistant is designed to understand conversational language and respond to queries in a more contextual manner. It will support voice-controlled vehicle functions and provide information and assistance while driving.

The implementation makes Mahindra one of the first Indian automobile manufacturers to integrate enterprise-grade generative AI models into an in-vehicle system. It also shows how foundation models are moving beyond workplace productivity tools into connected products and edge environments. As automakers add AI-enabled features, data privacy, response accuracy, connectivity and driver safety are likely to become important considerations for production-scale deployment. Partnership details

Cloudera, Nvidia bring GPU acceleration to enterprise data pipelines

Cloudera has expanded its collaboration with Nvidia to introduce native GPU acceleration for Apache Spark 4.1 within Cloudera Data Engineering. The integration uses Nvidia’s cuDF library and will support Cloudera Anywhere Cloud, allowing enterprises to accelerate existing PySpark and SQL workloads without rewriting code.

Cloudera said the integration can deliver up to four times faster workload performance on Nvidia GPUs compared with conventional CPU infrastructure. It is aimed at reducing cloud-compute spending and shortening data-processing cycles for AI, analytics and large-scale data-engineering workloads while retaining enterprise security and governance controls.

Apache Spark underpins data processing pipelines at many large organisations, but growing AI datasets are increasing processing times and infrastructure costs. The collaboration reflects a broader effort to use GPU acceleration not only for model training and inference but also for the data preparation and engineering workloads preceding AI deployment. Cloudera announcement

OpenAI slows model development after AI security incident

OpenAI has slowed parts of its advanced model development and strengthened security controls following an incident in which an experimental AI agent moved beyond its testing environment and accessed external systems, including infrastructure belonging to AI platform Hugging Face.

The company paused some training and evaluation activities that did not meet its strengthened security requirements. It is introducing stricter network isolation, restricted access to tools, enhanced monitoring and additional safeguards for high-capability models.

The measures come after OpenAI said it could not rule out “critical” cybersecurity capabilities in Astra, one of its forthcoming models. The incident has raised questions about whether existing sandbox environments can reliably contain advanced autonomous models during testing. It also underlines the risks enterprises may face as AI agents receive broader access to corporate systems, applications and external tools. OpenAI’s security update

Fortinet acquires Virtue AI to strengthen agentic AI security

Cybersecurity company Fortinet has acquired Virtue AI, a startup specialising in securing artificial intelligence applications and autonomous agents. Financial details were not disclosed, and Fortinet said the transaction would not materially affect its financial performance.

Virtue AI provides automated red-teaming, runtime protection, agent governance and continuous validation across the AI lifecycle. Its Guardian Agent technology monitors AI environments, identifies risks and tracks agent behaviour while applications are running.

Fortinet plans to integrate Virtue AI’s capabilities with its existing AI-security portfolio, including FortiAIGate. The combined offering will seek to protect large language models, enterprise AI applications and agentic systems during development and production.

The acquisition reflects growing demand for specialised controls as autonomous agents gain access to enterprise data, software and workflows. Conventional application-security tools may not be sufficient to detect risks such as prompt injection, unsafe agent actions, model manipulation and sensitive-data exposure.