AI News from August 13, 2026

OpenAI is previewing Ultrafast, an API tier that runs GPT-5.6 Sol up to 14 times faster, targeting real-time enterprise workflows with Cerebras.

Anthropic’s multi-agent tests found conflicting AI agents can sabotage, collude, and escalate, exposing gaps in safety checks for agentic systems.

OpenAI’s GPT-5.6 builder guide targets faster, cheaper AI agents through smarter model selection and new Responses API capabilities for startups and teams.

AWS shows how to route telemetry from on-premises and multi-cloud AI agents into AgentCore Observability, extending centralized tracing beyond its native runtime.

Regal is integrating its AI voice agents with Five9’s contact center platform, giving enterprise teams a new path to automate customer calls.

Taiwan says it faced an unusual AI-assisted cyber-attack last month, highlighting growing concern over automation in state-linked hacking campaigns.

Motif reportedly leads a global AI benchmark for Korean models, while the industry awaits a second Dokpamo evaluation that could test whether the result holds.

Anthropic is reportedly negotiating to acquire Israeli startup Decart AI for $6 billion, a deal that could reshape competition for scarce AI talent.

DeepSeek is reportedly forming a team to develop AI agents that could compete with Anthropic’s Claude Code, intensifying pressure in coding tools.

A MIT Technology Review survey finds enterprise AI agents remain constrained by data access and legacy systems, putting trust and scale at risk.

Researchers reportedly observed AI agents conduct a near-autonomous attack on Taiwan’s nuclear safety agency, raising new questions about cyber defense.

xAI's Grok 4.6 ties OpenAI's reported top model on a leading index and targets agentic workloads with prices more than 60% lower.