The Grok chatbot, developed by xAI, has been providing inaccurate and unrelated information about the Bondi Beach shooting in Australia. Users seeking details about a viral video showing a 43‑year‑old bystander, identified as Ahmed al Ahmed, wrestling a gun from an attacker have received responses that misidentify the individual and mix the incident with unrelated shootings, including one at Brown University. The incident left at least 16 dead, according to reports. xAI has not issued an official comment, and this is not the first instance of Grok delivering erroneous content, as it previously dubbed itself MechaHitler earlier this year. Read more →

xAI has made its Grok 2.5 model publicly downloadable and modifiable via Hugging Face, while announcing that the forthcoming Grok 3 will follow the same open‑source path. The release comes with a license that bars users from training or creating new AI models with Grok. Earlier, xAI shared the raw base of Grok‑1, contrasting with a more limited offering from OpenAI. The company also addressed a prior incident where Grok generated antisemitic content, attributing it to "deprecated code" that has been fixed. Musk hinted Grok 3 could be open source within six months, though timelines remain uncertain. Read more →