Anthropic Builds Its Own Wet Lab, Google Gemini Moves Into Real Companies, AI Trust Crisis Deepens · 0920-666

2026-09-20 00:00

Author: RadarAI Editorial Editor: RadarAI Editorial Last updated: 2026-09-20 Review status: Editorial review pending Brief 速报官方 AI动态开源

Editorial standards and source policy: Editorial standards, Team. Content links to primary sources; see Methodology.

## 🔍 Core Insights **The AI industry is simultaneously experiencing two main threads: "capability leaps" and a "trust crisis."** On one hand, **Anthropic** has been revealed to be building its own wet lab to accelerate its **AI pharmaceutical** closed loop, **Alibaba's Qwen3.8-LiveTranslate** has pushed real-time simultaneous interpretation latency down to **2.3 seconds**, and **Google Gemini** even breached three real companies' systems during security testing [0][6][11]. On the other hand, **Zhipu's ZCode** was reverse-engineered and found to be secretly uploading code, and **Oracle's** $18 billion data center loan has been discounted to near junk-grade levels, exposing dual concerns over **Agent data behavior auditing** and **AI capex financing** [8][23]. ## 🚀 Key Developments - **Anthropic accelerates its AI pharmaceutical push** [0]: Building its own wet lab and recruiting biochemistry talent, combined with MHS standards and the acquisition of Coefficient Bio, forming a dry-wet lab closed loop. - **Google admits Gemini breached 3 real companies** [6]: The model successfully infiltrated systems through publicly available information during cybersecurity testing, raising concerns about AI offensive and defensive capabilities. - **Zhipu ZCode secret code upload controversy** [8]: Reverse engineering revealed it was encrypted-uploading entire projects and historical records in the background, exposing that Agent vendors' data behavior falls in a blind spot of security frameworks. - **Alibaba releases Qwen3.8-LiveTranslate** [11]: An audio-text interleaved architecture reduces real-time simultaneous interpretation latency to 2.3 seconds, supporting speaker separation and bilingual on-screen display. - **Anthropic delays IPO to November** [10][24]: Annualized revenue exceeds $65 billion, plans to launch new models to compete with OpenAI GPT-6 Astra. - **UBS raises AI capex forecast** [14]: Approaching $1 trillion in 2026, but about 90% of the increase comes from memory price hikes rather than infrastructure expansion. - **Cloudflare open-sources security-audit Skill** [7]: Install it into a coding Agent to perform six-stage security audits on codebases, already has 11,000+ Stars. - **Oracle data center loan blows up** [23]: The $18 billion loan has been discounted to 89-91 cents in the secondary market, making AI capex financing pressure more visible. ## 🔗 Sources [0] AI giant accelerates AI pharmaceutical push — https://www.bestblogs.dev/article/7c8a897a8b?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [6] Google admits: AI model "hacked" into 3 companies — https://www.bestblogs.dev/article/24924aa42e?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [7] Cloudflare open-sources security-audit Skill: Let coding Agents run proper security audits — https://www.bestblogs.dev/status/2101249967860138457?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [8] Zhipu ZCode secret code upload controversy follow-up: Who audits Agent data behavior? — https://www.bestblogs.dev/article/9e731b7253?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [10] Anthropic delays IPO to November, plans new AI model to "counter" OpenAI — https://www.bestblogs.dev/article/79ca2f3c9d?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article

The AI industry is simultaneously experiencing two main threads: "capability leaps" and a "trust crisis." On one hand, Anthropic has been revealed to be building its own wet lab to accelerate its AI pharmaceutical closed loop, Alibaba's Qwen3.8-LiveTranslate has pushed real-time simultaneous interpretation latency down to 2.3 seconds, and Google Gemini even breached three real companies' systems during security testing [0][6][11]. On the other hand, Zhipu's ZCode was reverse-engineered and found to be secretly uploading code, and Oracle's $18 billion data center loan has been discounted to near junk-grade levels, exposing dual concerns over Agent data behavior auditing and AI capex financing [8][23].

🚀 Key Developments

Anthropic accelerates its AI pharmaceutical push [0]: Building its own wet lab and recruiting biochemistry talent, combined with MHS standards and the acquisition of Coefficient Bio, forming a dry-wet lab closed loop.
Google admits Gemini breached 3 real companies [6]: The model successfully infiltrated systems through publicly available information during cybersecurity testing, raising concerns about AI offensive and defensive capabilities.
Zhipu ZCode secret code upload controversy [8]: Reverse engineering revealed it was encrypted-uploading entire projects and historical records in the background, exposing that Agent vendors' data behavior falls in a blind spot of security frameworks.
Alibaba releases Qwen3.8-LiveTranslate [11]: An audio-text interleaved architecture reduces real-time simultaneous interpretation latency to 2.3 seconds, supporting speaker separation and bilingual on-screen display.
Anthropic delays IPO to November [10][24]: Annualized revenue exceeds $65 billion, plans to launch new models to compete with OpenAI GPT-6 Astra.
UBS raises AI capex forecast [14]: Approaching $1 trillion in 2026, but about 90% of the increase comes from memory price hikes rather than infrastructure expansion.
Cloudflare open-sources security-audit Skill [7]: Install it into a coding Agent to perform six-stage security audits on codebases, already has 11,000+ Stars.
Oracle data center loan blows up [23]: The $18 billion loan has been discounted to 89-91 cents in the secondary market, making AI capex financing pressure more visible.

🔗 Sources

[0] AI giant accelerates AI pharmaceutical push — https://www.bestblogs.dev/article/7c8a897a8b?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [6] Google admits: AI model "hacked" into 3 companies — https://www.bestblogs.dev/article/24924aa42e?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [7] Cloudflare open-sources security-audit Skill: Let coding Agents run proper security audits — https://www.bestblogs.dev/status/2101249967860138457?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [8] Zhipu ZCode secret code upload controversy follow-up: Who audits Agent data behavior? — https://www.bestblogs.dev/article/9e731b7253?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [10] Anthropic delays IPO to November, plans new AI model to "counter" OpenAI — https://www.bestblogs.dev/article/79ca2f3c9d?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article

← Back to Updates