Google Launches Gemini 3.5 Flash Cyber AI to Find and Fix Software Vulnerabilities 56%

By Ravie Lakshmanan12%

7/21/2026, 3:09:00 PM

BS Summary: This article contains 21 faulty reasoning types, including Optimism Bias, Attempt to Sell a Product or Service, and Framing Effect, with Appeal to Authority as the most egregious example at 34.1% saturation with 136 hits. Analysis detected 1,086 faulty-reasoning hits from 399 analyzed words, generating a BS Score of 53.2% and a BS Rank of 56% (9,521 of 21,202 articles). This article is worse (more manipulative) than 55.10% of the article peer group.

Google's DeepMind on Tuesday announced the release of Gemini 3.5 Flash Cyber, a specialized artificial intelligence (AI) model built atop 3.5 Flash that's designed to discover, validate, and patch vulnerabilities quickly and efficiently. 
According to the tech giant, the model will be exclusively available to governments and trusted partners via CodeMender as part of a limited-access pilot program. 
CodeMender is an AI-powered agent for vulnerability discovery and patching that was unveiled by the company in October 2025. 
A Google DeepMind spokesperson told The Hacker News that there are plans to extend the model's capabilities to include red-teaming features and end-to-end enterprise defense. 
The lightweight model, per DeepMind, is both cost-efficient and highly capable alternative to large, costly cybersecurity-focused models. 
CodeMender can call upon 3.5 Flash Cyber "multiple times at high speed and low cost," allowing the AI agent to scan more code paths and find vulnerabilities. 
The release of 3.5 Flash Cyber comes alongside Gemini 3.6 Flash and 3.5 Flash-Lite, which are optimized for improved coding, knowledge work, and multimodal performance and low-latency tasks, respectively. 
"Given the dual-use nature of this technology, we have taken an intentional approach to how we deploy 3.5 Flash Cyber," Raluca Ada Popa, DeepMind's Gemini Security Lead, and Four Flynn, vice president of security and privacy at DeepMind, said in a blog post shared with The Hacker News ahead of publication. 
"As part of a limited-access pilot program, 3.5 Flash Cyber will be exclusively available to governments and trusted partners via CodeMender, expanding over time. 
This will give frontline defenders a head start in finding and fixing critical vulnerabilities before they can be exploited, while mitigating against broader misuse." 
Like in the case of Anthropic and OpenAI, Google has put 3.5 Flash Cyber to the test to uncover remote code execution vulnerabilities in public APIs and a memory-corruption vulnerability in a sensitive production service. 
The model is also said to have produced a 100% reliable remote-code execution exploit that bypassed standard mitigation techniques like Address Space Layout Randomization (ASLR) and Write XOR Execute (W^X). 
Google said it's separately bringing CodeMender's foundational capabilities directly to customers with generally available Gemini models through the Gemini Enterprise Agent Platform. 
"By powering CodeMender with 3.5 Flash Cyber, we're providing a highly capable, scalable, and affordable architecture designed to help more defenders secure software," it added. 
Article reasoning-pattern comparisonThis article: 6.0%Ravie Lakshmanan: 1.5%The Hacker News: 2.0%Confirmation Bias6.0%This article: 0.0%Ravie Lakshmanan: 1.4%The Hacker News: 1.2%Anchoring Bias0.0%This article: 6.8%Ravie Lakshmanan: 2.2%The Hacker News: 3.3%Availability Heuristic6.8%This article: 0.0%Ravie Lakshmanan: 1.6%The Hacker News: 1.4%Representativeness Heuristic0.0%This article: 0.0%Ravie Lakshmanan: 1.0%The Hacker News: 0.6%Hindsight Bias0.0%This article: 11.8%Ravie Lakshmanan: 2.2%The Hacker News: 2.5%Overconfidence Bias11.8%This article: 23.3%Ravie Lakshmanan: 2.6%The Hacker News: 2.8%Framing Effect23.3%This article: 0.0%Ravie Lakshmanan: 0.9%The Hacker News: 1.1%Loss Aversion0.0%This article: 6.3%Ravie Lakshmanan: 0.4%The Hacker News: 0.6%Status Quo Bias6.3%This article: 0.0%Ravie Lakshmanan: 0.0%The Hacker News: 0.1%Sunk Cost Effect0.0%This article: 31.1%Ravie Lakshmanan: 1.2%The Hacker News: 1.3%Optimism Bias31.1%This article: 6.0%Ravie Lakshmanan: 1.3%The Hacker News: 1.6%Pessimism Bias6.0%This article: 7.5%Ravie Lakshmanan: 8.0%The Hacker News: 6.8%Negativity Bias7.5%This article: 0.0%Ravie Lakshmanan: 0.3%The Hacker News: 0.8%Self-Serving Bias0.0%This article: 0.0%Ravie Lakshmanan: 0.5%The Hacker News: 0.4%Fundamental Attribution Error0.0%This article: 0.0%Ravie Lakshmanan: 0.1%The Hacker News: 0.1%Actor-Observer Bias0.0%This article: 0.0%Ravie Lakshmanan: 0.1%The Hacker News: 0.1%In-Group Bias0.0%This article: 0.0%Ravie Lakshmanan: 1.4%The Hacker News: 0.4%Out-Group Homogeneity Bias0.0%This article: 10.5%Ravie Lakshmanan: 0.4%The Hacker News: 0.6%Halo Effect10.5%This article: 0.0%Ravie Lakshmanan: 0.0%The Hacker News: 0.0%Horn Effect0.0%This article: 0.0%Ravie Lakshmanan: 0.1%The Hacker News: 0.0%Dunning-Kruger Effect0.0%This article: 8.8%Ravie Lakshmanan: 2.2%The Hacker News: 1.4%Recency Bias8.8%This article: 0.0%Ravie Lakshmanan: 0.3%The Hacker News: 0.2%Primacy Effect0.0%This article: 0.0%Ravie Lakshmanan: 0.1%The Hacker News: 0.1%Blind-Spot Bias0.0%This article: 0.0%Ravie Lakshmanan: 0.2%The Hacker News: 0.1%Ad Hominem0.0%This article: 0.0%Ravie Lakshmanan: 0.0%The Hacker News: 0.1%Straw Man0.0%This article: 34.1%Ravie Lakshmanan: 4.8%The Hacker News: 4.0%Appeal to Authority34.1%This article: 6.0%Ravie Lakshmanan: 0.4%The Hacker News: 1.6%False Dilemma6.0%This article: 0.0%Ravie Lakshmanan: 0.3%The Hacker News: 0.5%Slippery Slope0.0%This article: 0.0%Ravie Lakshmanan: 0.0%The Hacker News: 0.1%Circular Reasoning0.0%This article: 11.8%Ravie Lakshmanan: 3.8%The Hacker News: 4.3%Hasty Generalization11.8%This article: 0.0%Ravie Lakshmanan: 0.2%The Hacker News: 0.1%Red Herring0.0%This article: 0.0%Ravie Lakshmanan: 0.2%The Hacker News: 0.2%Bandwagon0.0%This article: 0.0%Ravie Lakshmanan: 1.2%The Hacker News: 1.1%Appeal to Emotion0.0%This article: 0.0%Ravie Lakshmanan: 0.2%The Hacker News: 0.5%Begging the Question0.0%This article: 6.0%Ravie Lakshmanan: 2.1%The Hacker News: 1.9%Post Hoc (False Cause)6.0%This article: 0.0%Ravie Lakshmanan: 0.0%The Hacker News: 0.0%Tu Quoque0.0%This article: 7.5%Ravie Lakshmanan: 1.0%The Hacker News: 0.6%Burden of Proof7.5%This article: 0.0%Ravie Lakshmanan: 0.1%The Hacker News: 0.1%Appeal to Nature0.0%This article: 0.0%Ravie Lakshmanan: 0.0%The Hacker News: 0.3%Composition/Division0.0%This article: 0.0%Ravie Lakshmanan: 0.3%The Hacker News: 1.0%Anecdotal0.0%This article: 0.0%Ravie Lakshmanan: 0.1%The Hacker News: 0.1%No True Scotsman0.0%This article: 3.3%Ravie Lakshmanan: 4.0%The Hacker News: 2.3%Ambiguity (Equivocation)3.3%This article: 0.0%Ravie Lakshmanan: 0.0%The Hacker News: 0.0%Gambler’s Fallacy0.0%This article: 0.0%Ravie Lakshmanan: 0.0%The Hacker News: 0.0%Middle Ground0.0%This article: 0.0%Ravie Lakshmanan: 0.0%The Hacker News: 0.0%Personal Incredulity0.0%This article: 0.0%Ravie Lakshmanan: 0.0%The Hacker News: 0.1%Special Pleading0.0%This article: 0.0%Ravie Lakshmanan: 0.0%The Hacker News: 0.1%Genetic Fallacy0.0%This article: 15.8%Ravie Lakshmanan: 3.4%The Hacker News: 1.4%Unattributed Quote15.8%This article: 19.0%Ravie Lakshmanan: 2.8%The Hacker News: 1.0%Quote-first Misdirection19.0%This article: 16.0%Ravie Lakshmanan: 2.4%The Hacker News: 2.3%Biased Writer Voice16.0%This article: 6.0%Ravie Lakshmanan: 2.5%The Hacker News: 4.5%Indoctrination6.0%This article: 0.0%Ravie Lakshmanan: 0.0%The Hacker News: 0.0%Politically Left Leaning Bias0.0%This article: 0.0%Ravie Lakshmanan: 0.0%The Hacker News: 0.0%Politically Right Leaning Bias0.0%This article: 28.6%Ravie Lakshmanan: 0.5%The Hacker News: 2.9%Attempt to Sell a Product or S…28.6%

399 words analyzed.

Speakers

1speaker6.3%attributed speech374writer words
Voice mapSelect a segment to jump to its words
Writer's voice • 13 words • 100.0% coverageWriter's voice • 33 words • 100.0% coverageWriter's voice • 25 words • 100.0% coverageWriter's voice • 19 words • 0.0% coverageThe Hacker News • 25 words • 0.0% coverageWriter's voice • 17 words • 0.0% coverageWriter's voice • 27 words • 0.0% coverageWriter's voice • 29 words • 100.0% coverageWriter's voice • 51 words • 100.0% coverageWriter's voice • 24 words • 0.0% coverageWriter's voice • 24 words • 100.0% coverageWriter's voice • 35 words • 0.0% coverageWriter's voice • 30 words • 100.0% coverageWriter's voice • 22 words • 100.0% coverageWriter's voice • 25 words • 100.0% coverage
Selected voice

The Hacker News

100%flagged-word coverage
25 attributed words100% of attributed speech95% writer coverage
0%17.5%35.0%Attempt to Sell a Product -30.5 ptsWriter: 30.5%The Hacker News: 0.0%0.0%Quote-first Misdirection-20.3 ptsWriter: 20.3%The Hacker News: 0.0%0.0%Biased Writer Voice-17.1 ptsWriter: 17.1%The Hacker News: 0.0%0.0%Unattributed Quote-16.8 ptsWriter: 16.8%The Hacker News: 0.0%0.0%Indoctrination-6.4 ptsWriter: 6.4%The Hacker News: 0.0%0.0%

Attribution is sentence-level. Pattern percentages are calculated only from words assigned to that voice.

Loading…
Loading…
Loading…

Analysis

Hover over highlighted words in the article to view the associated bias or fallacy analysis.