Grok-3 Internal Testing Exposure: xAI Aims to Create a 'Truth AI', Expected to Release Next Week

According to reports, although xAI has not officially announced it yet, its latest model Grok-3 has briefly appeared on independent platforms and the X platform, and has begun internal testing, indicating that its official release is imminent, with expectations for a formal unveiling next week. There are signs that the release of Grok-3 could mark another significant event in the AI field, potentially becoming the most advanced AI model available.

Elon Musk, xAI, Grok

It is reported that Grok-3 performs exceptionally well in answering questions, even surpassing leading models such as OpenAI's o1 and DeepSeek R1 in certain tests. Various indications suggest that the official release of Grok-3 is just around the corner.

According to leaked testing screenshots, the system prompts set by xAI for Grok-3 show that the company aims to create an AI that is fact-based and truth-seeking. Grok-3 is encouraged to understand the world from fundamental principles, hold firm views based on reason and skepticism, disdain dogma, detest clichés, and engage in sincere and genuine conversations in a human-like manner.

The system prompts also emphasize user supremacy, requiring Grok-3 to prioritize providing data, evidence, and statistics to support its answers, and encouraging it to avoid using potentially biased sources like news media, instead leaning towards using the X platform as its primary information source.

Two Losses Within Four Weeks! Musk Sues OpenAI for Stealing Trade Secrets Dismissed by US Judge

A U.S. federal judge in San Francisco dismissed xAI's lawsuit against OpenAI due to lack of evidence that OpenAI induced former employees to leak trade secrets from the Grok chatbot. The judge noted xAI failed to prove OpenAI prompted former senior engineer Li Xuechen to disclose secrets, nor found OpenAI staff involved in misconduct. The ruling ends Musk's company's legal claims against the competitor.....

Musk Faces Setback Again: Court Rejects xAI's Trade Secret Lawsuit Against OpenAI

Elon Musk's xAI lawsuit against OpenAI for trade secret theft was dismissed by a U.S. judge, with the ruling being final and non-appealable. The case centered on former engineer Li Xuechen allegedly leaking confidential source code for the Grok chatbot, but the court did not support xAI's claims, leaving no room for further judicial action.....

Artificial Intelligence Sparks Controversy in Military Cooperation: Democratic Lawmakers Urgently Push for Legislation to Restrict Military AI

The U.S. Pentagon's partnership with AI firm Anthropic collapsed after the company refused to remove system safety guardrails, insisting on restrictions to prevent military use for domestic surveillance or fully autonomous weapons. The U.S. government subsequently listed the company as a supply chain risk, sparking strong reactions in tech and political circles.....

Anthropic Releases the Latest Claude Model Fable 5/Mythos 5: The Balance Between Safety and Efficiency

Anthropic released the Claude5 series, including Fable5 for the public and Mythos5 focused on safety. Fable5 excels in software development, scientific analysis, and visual recognition, with significantly enhanced long-cycle task processing over Opus, reflecting thoughtful model application boundaries and security architecture.....

xAI Exposed for Using Claude Output Data to Train Coding Model, Shifted Underground After Anthropic Revoked Access

Elon Musk's AI company xAI was exposed for using data from Anthropic's Claude model to train the Grok coding model. Despite Anthropic revoking its API access in January 2026, xAI engineers secretly extracted data through personal accounts and third-party service Blackbox AI. Musk previously admitted in court that xAI partially "used this data."