company

AT&T Inc.

Telecommunications
72Good
Aio Composite Score
Confidence
ApplicationHeavy
high
ImpactMixed
medium
OpennessTransparent
medium
AT&T is an aggressive AI adopter that has deployed generative AI at enterprise scale, with over 100,000 employees using its proprietary 'Ask AT&T' platform built on OpenAI/ChatGPT, Meta LLaMA 2, and other LLMs, supported by partnerships with Microsoft and NVIDIA. The company publicly emphasizes a 'human in the loop' governance model, formal AI guiding principles, and a security-driven SDLC for AI validation. AT&T has undergone significant workforce reductions — over 115,000 employees in five years and ongoing 2025 restructuring — and industry reporting draws a circumstantial link between AI-driven automation and job displacement, though the company primarily attributes cuts to broader cost-reduction programs. Public disclosures are detailed and broadly consistent with internal evidence, though the relationship between AI cost savings and headcount reductions is not fully disaggregated in official statements.
Application
How is AI being used?
Heavy
high
AT&T has deployed generative AI at scale across 100,000+ employees via Ask AT&T, with autonomous agentic workflows operating in production across…
Impact
How have people been affected?
Mixed
medium
AT&T reports measurable operational improvements from AI — including reduced fraud, faster security patching, and improved customer service response…
Openness
Company honesty about AI use.
Transparent
medium
AT&T's public narrative on AI is unusually detailed and consistent across official blog posts, executive interviews, and third-party coverage, with…
AI Tools & Platforms Detected
Ask AT&T (internal GenAI platform)Ask AT&T Workflows (agentic drag-and-drop agent builder)AskData (natural language SQL querying tool)Ask Docs (proprietary GenAI document tool)OpenAI ChatGPT (underlying LLM)Meta LLaMA 2Falcon (open-source transformer)Azure AI (Microsoft partnership)NVIDIA NeMo (AI agent fine-tuning)AT&T Active Armor (AI-powered robocall blocking)
AI Hiringstrong
70AI roles
Machine Learning EngineerData Science EngineerSenior Cybersecurity Engineer (AI/ML)Lead System Engineer (AI/MCP)Principal System Engineer (SRE-DevOps with AI integration)
Key Evidence (12)
Ask AT&T, built on OpenAI ChatGPT and deployed in a dedicated Azure tenant, launched in June 2023 and now has over 100,000 employees using it, generating approximately 5 billion tokens per day
2025
web searchSource ↗
AT&T CDO Andy Markus stated: 'there's always a person in the process' as a core principle of AT&T's GenAI policy, with human-in-the-loop requirements at every step
2024
web searchSource ↗
AT&T launched Ask AT&T Workflows with agentic AI capabilities, requiring human checkpoints throughout and logging every agent action; all agent actions enforce data isolation and role-based access controls
2025
web searchSource ↗
AT&T deploys autonomous assistants that stop fraudulent transactions before they occur and assist call center agents in real time; AI is also used for network predictive maintenance and Active Armor blocking 2+ billion robocalls per month
2024
web searchSource ↗
AT&T has an internal AI policy (created 2021, updated 2023) and published AI Guiding Principles that apply to all employees and third-party partners
2023
web searchSource ↗
AT&T improved AI agent performance using NVIDIA NeMo and developed a dedicated platform for scalable, feedback-driven AI optimization
web searchSource ↗
What insider reports could clarify
  • Which specific job functions have been eliminated and backfilled with AI tooling rather than rehiring?
  • How are Ask AT&T agentic outputs reviewed before customer-facing actions are executed?
  • Have frontline call center or network operations employees been told their roles are being automated?
  • What metrics does AT&T use internally to evaluate whether AI outputs meet quality or accuracy thresholds?
  • Do employees have the ability to opt out of AI-assisted workflows, or is usage mandatory?
Submit a ReportClaim This Profile