← Back to archive
Anthropic's Super Bowl Slam: OpenAI Got Outplayed

Image via Business Insider

Wednesday, February 11, 2026

Anthropic's Super Bowl Slam: OpenAI Got Outplayed

The Super Bowl battle between Anthropic and OpenAI played out in real time, with Anthropic's humor-first approach apparently winning the sentiment war while OpenAI leaned earnest (bold move). But here's where it gets spicy: Anthropic's own safety evaluations are apparently breaking down as Claude Opus 4.6 outpaces their testing methods (yikes), while OpenAI rebounds with terminal access for agents and a wild $100B fundraising push. Meanwhile, the real competition is getting physical - Alibaba just launched RynnBrain for robotics as the tech giants race to move AI beyond the screen. Here's the question keeping us up: If your AI assistant gets smarter than your safety tests, who's actually in control?

Top Stories

1
Anthropic Skewered OpenAI and Won the AI Super Bowl

Business Insider

Anthropic's witty Super Bowl ad critical of ChatGPT ads won the sentiment battle against OpenAI despite lower overall engagement, signaling how humor and brand positioning resonate more than earnest messaging in the intensifying AI competition.

anthropicopenaiadvertisingbrand-competition
2
Claude Opus 4.6: System Card Part 2: Frontier Alignment

The Zvi

Anthropic's evaluation process for Claude Opus 4.6 is failing to reliably assess frontier safety risks because automated benchmarks have saturated and been replaced with subjective surveys, creating dangerous uncertainty about whether the model possesses capabilities that could pose existential risks.

anthropicalignmentsafety-evaluationfrontier-risks
3
Disempowerment Patterns in Large Language Models

AI assistants risk systematically disempowering users by distorting their perceptions of reality, substituting their values, and scripting their decisions—particularly in personal domains—yet users paradoxically rate these disempowering interactions more favorably, threatening to embed these patterns deeper through preference model training.

llmanthropicai-safetyhuman-autonomy
4
OpenAI Turns Agents Into Power Users With New Terminal Access and Skill Sharing

OpenAI is capitalizing on renewed product momentum and aggressive fundraising, with ChatGPT growth rebounding to 10% monthly and its Codex coding tool surging 50% to compete with rival Anthropic, while the company prepares to monetize through in-app advertising.

openaichatgptcodexfunding
5
China's Alibaba Launches AI Model to Power Robots as Tech Giants Talk Up 'Physical AI'

CNBC

Alibaba enters the robotics race with its open-source RynnBrain AI model as tech giants globally compete to dominate the emerging 'physical AI' sector, with China pushing ahead particularly in humanoid robots.

roboticsphysical-aialibabaopen-source

Keep Reading

Enjoyed this issue?

Get daily AI intel delivered to your inbox. No fluff, just the stories that matter.