Anthropic's AI Coding Tool Leaks Its Own Source Code for the Second Time

Anthropic is facing renewed scrutiny after its AI coding tool accidentally leaked its own source code for the second time within a year. A researcher flagged the critical vulnerability on March 31, 2026, and the exposed code has since been quickly archived across multiple public repositories. This recurring security lapse raises fresh questions about the company's software release practices, internal protocols, and the overall safety of its proprietary tools in an increasingly competitive AI landscape.

አንትሮፒክ (Anthropic) የኤአይ የኮዲንግ መሳሪያው በአንድ ዓመት ውስጥ ለሁለተኛ ጊዜ የራሱን ሚስጥራዊ ኮድ (source code) በስህተት በማጋለጡ ምክንያት ከባድ ትችት እና ምርመራ እየገጠመው ይገኛል። አንድ ተመራማሪ በማርች 31 ቀን 2026 ይህንን የደህንነት ክፍተት ያጋለጠ ሲሆን፣ ሾልኮ የወጣው ኮድ በተለያዩ የህዝብ ማከማቻዎች (public repositories) ላይ በፍጥነት ተሰራጭቷል። ይህ ተደጋጋሚ የደህንነት ችግር፣ በኩባንያው የሶፍትዌር አቀራረብ ስርዓት፣ የውስጥ ቁጥጥር፣ እና ውድድር በበዛበት የሰው ሰራሽ አስተውሎት ገበያ ውስጥ የቴክኖሎጂው አስተማማኝነት ላይ አዳዲስ ጥያቄዎችን አስነስቷል።

@webthreeth

yeah so guess what, I'm definitely not paying for Claude Max. I'll just use GLM 5.1 or Gemini 3.1 Pro in Claude Code instead. nothing beats leaked and open source stuff.

​https://github.com/Gitlawb/openclaude

Done Installing.

Pro tip for anyone in Ethiopia or other regions where Claude Code is blocked. You won't need to use a VPN at all. This works perfectly right out of the box and it's totally free

Check out this Hackton.

https://lablab.ai/ai-hackathons/amd-developer?enroll=true

Made some investments today. Which one should I start with?

Is it bad if i wash a popcorn and put it in wet before it dries? Did I mess up on this? 😭

Fasika vibes as a developer 😍

been messing around with the new glm 5.1 model and found out you can get free access to it at https://modal.com/

@snack_shell

Anthropic Unveils Claude Opus 4.7 with Enhanced Vision and Autonomous Capabilities

Anthropic has announced the release of Claude Opus 4.7, its most advanced Opus model to date. The upgraded AI delivers improved precision in following instructions, handles long-running tasks with greater rigor, and autonomously verifies its own outputs before reporting back. A major highlight is its significantly enhanced vision capability, which processes images at more than three times the previous resolution to generate high-quality interfaces, slides, and documents.

አንትሮፒክ (Anthropic) እስካሁን ካበለጸጋቸው የ"ኦፐስ" (Opus) ሞዴሎች እጅግ የላቀ አቅም ያለውን "ክሎድ ኦፐስ 4.7" (Claude Opus 4.7) ይፋ አድርጓል። ይህ የተሻሻለ ሞዴል ትዕዛዞችን በከፍተኛ ትክክለኛነት የሚፈጽም፣ ረጅም ጊዜ የሚወስዱ ተግባራትን በጥልቀት የሚያከናውን እና ምላሽ ከመስጠቱ በፊት የራሱን ውጤቶች አስቀድሞ የሚያረጋግጥ ነው። በተለይም የእይታ (vision) አቅሙ በከፍተኛ ሁኔታ የተሻሻለ ሲሆን፣ ምስሎችን ከቀድሞው ከሶስት እጥፍ በላይ በሆነ ጥራት መተንተን በመቻሉ፣ የተሻሉ የበይነገጽ (interfaces)፣ የስላይድ እና የሰነድ ዝግጅቶችን ያቀርባል።

@webthreeth