Skip to main content

Research

Summaries and analysis of studies, standards, and benchmarks that matter with AI.

A collection of 27 posts

The assistant, not the author.
Education

The assistant, not the author.

How AI is engraining in New England schools: From the tools to PD to the pilots.

· 4Fsh Team

RAG Poisoning and the Drift You Don’t See
Security

RAG Poisoning and the Drift You Don’t See

Change the retrieval, change the answer. RAG turns inputs into control channels. Plant a sentence and the assistant will show it with confidence.

· 4Fsh Team

Teacher Training, Two Playbooks
Education

Teacher Training, Two Playbooks

From Rhode Island to NYC, states steer AI with PD and pilots. China mandates hours and competencies. Two playbooks. Same classroom stakes.

· 4Fsh Team

Writing AI Into School
Education

Writing AI Into School

China mandates AI class hours; the Northeast integrates with guardrails. What that means for teachers' workload, planning, and assessment.

· Sagger Khraishi

China & the US Northeast are setting AI’s Ground Rules for K-12
Education

China & the US Northeast are setting AI’s Ground Rules for K-12

How New York, Massachusetts, Rhode Island, Maine, and New Hampshire are shifting to teacher-led AI use - while China mandates hours and scope.

· Sagger Khraishi

Feel like you write like an AI? You kind of are...
Research

Feel like you write like an AI? You kind of are...

If you use AI to write the first draft, and then edit... There's bad news - backed by science.

· Sagger Khraishi

Why AI Just Can't Fix that Bug
Research

Why AI Just Can't Fix that Bug

The steps between feeling like a Prompting God to feeling like you are being gas lit by an AI is roughly 7 steps.

· Sagger Khraishi

AI generated Review Floods
Security

AI generated Review Floods

Review floods are a reputational DoS. How AI-made fakes, extortion, and pile-ons drown trust–and what founders can do to stay resilient.

· Sagger Khraishi

High-Confidence Wrong: AI and Insecure Code
Security

High-Confidence Wrong: AI and Insecure Code

Studies show ~45% of AI-generated code carries OWASP-class flaws. The threat isn’t exotic - it’s ordinary defaults shipped at machine speed.

· Sagger Khraishi

The Trojan Stack: Fake AI APIs and SDKs
Security

The Trojan Stack: Fake AI APIs and SDKs

Malicious AI-branded SDKs, fake APIs, and extensions are the new supply chain. How they steal tokens and hurt startups - and what to watch.

· Sagger Khraishi

Industrialized AI Intimacy Scams
Security

Industrialized AI Intimacy Scams

Inside shift-based, AI-oiled chat factories that scale intimacy across apps and convert attention into deposits.

· Sagger Khraishi

When a Meeting Isn’t a Meeting
Security

When a Meeting Isn’t a Meeting

Live deepfake calls blend faces, voices, and authority to move money. Arup's loss shows founders must treat meetings as attack surfaces.

· Sagger Khraishi

The Clone On The Line
Security

The Clone On The Line

AI-cloned voices turn routine calls into urgent traps. How vishing exploits trust, scales with kits, and targets SMBs.

· Sagger Khraishi

AI Catfishing at Scale: How SMBs Get Hooked
Security

AI Catfishing at Scale: How SMBs Get Hooked

Catfishing has gone corporate. IMFI uses LLMs, scripts, and handoffs to mirror founders and funnel affection into theft.

· Sagger Khraishi

When “Prove-It” Stops Proving Anything
Security

When “Prove-It” Stops Proving Anything

AI makes voices, faces, and behavioral biometrics easy to counterfeit. Identity checks become theater. Assurance erodes.

· Sagger Khraishi

AI Phishing Kits Sound Like Your Boss
Security

AI Phishing Kits Sound Like Your Boss

AI phishing kits mimic executives, sync email, chat, and SMS, and regenerate when blocked. The scam didn't change–the scale and polish did.

· Sagger Khraishi

Deepfake-Driven Fraud: The Breach Is Belief
Security

Deepfake-Driven Fraud: The Breach Is Belief

Arup lost $25M to a deepfake meeting. FBI warns AI voice/video scams are surging. The real breach isn’t code - it's belief

· Sagger Khraishi

The paper trail you can’t see
Security

The paper trail you can’t see

SMBs face realistic PDFs and QR-code phishing that bypass filters and MFA. How the invoice con works - and why scanning isn’t safer.

· Sagger Khraishi

Synthetic Identity & The Liveness Illusion
Security

Synthetic Identity & The Liveness Illusion

Synthetic identity fraud is exploding as gen-AI overlays bypass liveness; virtual camera injection and face swaps surged in 2024-2025.

· 4Fsh Team

The only honest way to measure AI success
Research

The only honest way to measure AI success

Use a five-layer scorecard - outcomes, adoption, quality & safety, cost, and risk to measure AI with ISO/NIST impact. Updated for 2025.

· 4Fsh Team

Just Say "No": How AI Politeness Drives up Costs
Research

Just Say "No": How AI Politeness Drives up Costs

There is a Dollar cost to Token Outputs. Which means when training your Company's AI - there is a fine line between polite and cost efficient.

· Sagger Khraishi

Copycat Apps: The impact on Innovation, AI Saturation, and Prompts
Strategy

Copycat Apps: The impact on Innovation, AI Saturation, and Prompts

Explore how copycat apps from mobile games to AI tools impact innovation, and economics - and why prompts are never going to go away as a result.

· 4Fsh Team

How LLM’s Downplay Intelligence to Match Roles
Research

How LLM’s Downplay Intelligence to Match Roles

A research analysis summary about how AI can change it's intelligence to fit it's role.

· 4Fsh Team

Ethical AI Publishing: Navigating the Future of Scientific Research
Research

Ethical AI Publishing: Navigating the Future of Scientific Research

Learn more about how AI is reshaping the future of scientific publishing, manuscript reviews, research discoverability, and best practices.

· 4Fsh Team

How Pre-training boosts AI performance
Research

How Pre-training boosts AI performance

Pre-training is where models learn. See how data quality, MoE, and long-sequence training drive real-world gains - and why fine-tuning can’t fix a weak base.

· Sagger Khraishi

Climate Modeling & AI in Environmental Research
Research

Climate Modeling & AI in Environmental Research

AI is upgrading climate modeling with improving accuracy, predicting extreme weather, and guiding environmental action. Updated for 2024

· 4Fsh Team

The Fascinating Evolution of AI Prompting
Research

The Fascinating Evolution of AI Prompting

Dive into the fascinating world of AI prompting, its evolution, and how it's shaping our interactions with artificial intelligence.

· Sagger Khraishi