510 summarised stories in Safety & Ethics, each linking back to the original source. Browse all topics →
OpenAI 2 years ago 31 ● 3 sources
OpenAI disrupted a Russia-linked operation called Bad Grammar that used AI to generate comments in English and Russian on Telegram targeting Ukraine, Moldova, Baltic states, and US politics. The operation generated comments across multiple geopolitical topics to influence discussions on the platform. Removing the accounts limits the operation's ability to conduct coordinated inauthentic behaviour at scale.
OpenAI 2 years ago 4 ● 3 sources
OpenAI shut down accounts linked to China's "Spamouflage" operation, which had been using AI tools to study social media patterns, create posts, and test a previously undisclosed website. The operation used ChatGPT and related services to support coordinated inauthentic behavior across multiple platforms. The ban prevents further development of this AI-enabled influence infrastructure while signaling enforcement action against state-linked information operations.
OpenAI 2 years ago 36 ● 3 sources
OpenAI banned accounts linked to a Russian influence operation called Doppelganger that used AI to generate anti-Ukraine content across social media and websites in multiple languages. The operation created thousands of coordinated social media accounts to spread misleading narratives about the Ukraine conflict. This action removes a significant source of automated disinformation targeting Ukraine during an ongoing geopolitical conflict.
OpenAI 2 years ago 24 ● 2 sources
OpenAI removed accounts associated with an Israeli operation called Zero Zeno that used AI systems to generate coordinated inauthentic content supporting Israel and Hamas opponents across multiple platforms. The operation generated content targeting anti-Hamas narratives, anti-Qatar messaging, pro-Israel positions, anti-BJP positions, and pro-Histadrut labor union support. The enforcement action demonstrates how state-linked actors are using AI tools for influence campaigns and illustrates platform efforts to detect and remove coordinated inauthentic behavior at scale.
OpenAI 2 years ago 33 ● 2 sources
OpenAI removed accounts linked to IUVM, an Iran-based operation that used AI to generate and translate content promoting Iran while criticizing Israel and the US across websites. The operation leveraged AI models to scale propaganda content across multiple platforms and languages. The takedown reduces one coordinated inauthentic behavior network's ability to distribute state-sponsored messaging at scale.
The daily briefing
TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.
TLDRocket
8 August 2026
Mistral AI's release of Shieldstral 1.0 3B marks a practical inflection point in content moderation: safety classifiers are shedding their fixed rulebooks. The 3B model takes policies as plain-language prompts at inference time—meaning a single deployment can enforce different standards for different customers without retraining—while matching the performance of a 20B baseline on both text and multimodal content. It runs on a single 16GB GPU and ships under Apache 2.0, making it available for local deployment across vLLM and llama.cpp. For teams managing multiple content policies or operating across jurisdictions, this matters because moderation no longer requires separate model forks or constant retraining cycles.
Read the full briefing →