AI Phishing Detection  ·  Private Beta

Something looks off.
Get a verdict
in seconds.

Suspicious email or text message? PhishEd Intelligence analyzes it end-to-end — reading images, decoding QR codes, and inspecting links live — then replies with a verdict, risk score, and exactly what to do next.

Be first to access when we launch

Private beta launching soon  ·  No spam

Works with
Gmail Outlook Apple Mail Proton Mail Yahoo Mail SMS/MMS screenshot

Dear Valued Customer,

We've detected unusual sign-in activity on your account. Your access has been temporarily limited for your protection.

Verify your identity within 24 hours to avoid permanent suspension:

secure-paypa1.com/verify/account

PE
PhishEd Intelligence — Analysis
Verdict
Likely Phishing
Risk Score
87/100

Why we flagged it
"secure-paypa1.com" spoofs PayPal using a lookalike character
Sandbox confirmed redirect to a credential-harvesting login clone
Urgent threatening language designed to pressure immediate action

Recommended actions
1Do not click any links or enter credentials on linked pages
2Report as phishing in your email client and delete the message
<30s
Average analysis time
3 verdicts
Phishing · Suspicious · Safe
5 free
Scans per 24 hours
Zero setup
No account or app required
2-stage AI
Fast model + deep model fallback

Phishing is the #1 cause of data breaches.
Most people have no way to check.

Attackers have industrialized phishing. The defenses haven't kept up — especially for the organizations that can't afford enterprise security tools.

91%
of cyberattacks begin with a phishing email
Deloitte
$4.76M
average cost of a data breach caused by phishing
IBM Security
3.4B
phishing emails sent every single day
AAG IT Support
1in 3
employees click a phishing link without security training
Proofpoint

From forward to verdict in seconds

No account. No app. No setup. Forward anything suspicious and PhishEd Intelligence replies with a full analysis — right back to your inbox.

1

Submit anything suspicious

Forward an email, attach a screenshot, include a QR code photo, or paste suspicious text. Any combination accepted — no formatting required.

2

Content extraction

The full message is parsed — sender identity, subject, all embedded links, normalized domains, and every attachment are extracted for downstream analysis.

3

OCR & image analysis

Image attachments are scanned with OCR. QR codes are decoded and their destinations inspected. Image-only phishing — built to evade text filters — is analyzed just like any other message.

4

Threat intelligence checks

Every link and domain is cross-referenced against Google Safe Browsing, VirusTotal, and real-time threat databases. Email authentication (SPF, DKIM, DMARC) is validated. Newly registered domains are flagged. When signals are conclusive, a verdict is issued without AI — saving time and improving accuracy.

5

Sandbox URL inspection

Suspicious URLs are opened in an isolated browser to observe the actual destination — redirect chains, harvesting forms, and deceptive login clones surface without any risk to you.

6

Verdict delivered to your inbox

A fast AI screens every submission. Low confidence escalates to a deeper model. You receive a verdict, risk score, specific reasons, and recommended next steps.

What you can submit
Email, SMS, messaging apps, and more
Forward an email
No login, no app — just forward
Beta
Screenshot or image
Attach any screenshot — OCR reads the text
Beta
QR code photo
Decoded and destination URL analyzed
Beta
SMS / MMS
Text a screenshot directly for smishing analysis
Soon
WhatsApp & iMessage
Forward suspicious chats for analysis
Soon
Social media DMs
Instagram, LinkedIn, Twitter/X scam messages
Soon
Web dashboard
Submit via browser, view history
Soon
REST API
Integrate into SIEM or security toolchain
Soon

Built for how phishing actually works today

Modern attacks use image-based evasion, QR obfuscation, and multi-hop redirects to bypass conventional filters. PhishEd Intelligence is built to catch them.

Links & Domains

Malicious URL detection

Every link is extracted and cross-checked against multiple threat feeds. Lookalike domains, newly registered hostnames, and multi-hop redirect chains are all flagged.

OCR

Image-based phishing

Optical character recognition reads text inside image attachments — catching scams designed to hide content from text-based filters entirely.

QR Codes

Quishing detection

QR codes in images are decoded and their destination URLs analyzed — a fast-growing vector used to sneak malicious links past email scanners.

SMS

Smishing analysis

Suspicious text messages forwarded as screenshots run through the same full pipeline as email — links, images, and sender metadata all analyzed.

Attachments

Fake invoices & documents

Image and PDF attachments are analyzed via OCR and their SHA256 hashes checked against VirusTotal's malware database — catching reused malicious payloads instantly.

Sender

Spoofing & impersonation

Sender headers, display names, and lookalike domains are checked for brand impersonation and executive fraud. SPF, DKIM, and DMARC are validated to catch authentication failures attackers rely on.

Sandbox

Live URL inspection

Uncertain URLs are opened in an isolated browser. Redirect chains, harvesting forms, and cloaked payloads are exposed without risk to you or your device.

AI

Two-stage AI analysis

A fast model screens every submission. When confidence is low or signals are mixed, a deeper reasoning model takes over for a more thorough verdict.

Threat Intelligence

Multi-feed reputation checks

URLs and IPs are cross-referenced against Google Safe Browsing, VirusTotal, and multiple real-time threat databases — independent sources running in parallel before AI is involved.

Email Authentication

SPF, DKIM & DMARC validation

Every submission is checked against the sender domain's published authentication records. A failing SPF or DMARC combined with other signals is a strong indicator of spoofing — and we weight it accordingly.

Security for the teams that need it most

Enterprise email security tools were built for Fortune 500 IT departments. PhishEd Intelligence was built for everyone they ignored.

Small & mid-sized businesses

Constantly targeted, no dedicated security team. Zero-setup means no IT involvement — just share one email address with your staff.

Law & accounting firms

Among the highest-value BEC targets — a single wire fraud can cost six figures. Professional services firms need protection without enterprise overhead.

Schools & districts

Staff are targeted with fake IT, vendor, and payroll emails. District-wide coverage requires nothing — just share the address. No software, no training.

Healthcare practices

HIPAA compliance pressure, high phishing rates, and minimal IT resources make small practices especially vulnerable — and underserved by enterprise tools.

Nonprofits & charities

Regularly targeted for donor data and wire transfer fraud. Operating on tight budgets, with staff who need clear guidance — not just a block notification.

Remote workers & freelancers

Personal and work email blend together — but employer security tools only cover corporate accounts. PhishEd works on any inbox, on any device.

Enterprise protection. Without the enterprise contract.

Enterprise tools like Proofpoint and Mimecast are built for large IT departments. PhishEd Intelligence delivers the same depth of analysis — with zero deployment, zero seats minimum, and a fraction of the cost.

PhishEd Intelligence Enterprise Tools
Proofpoint, Mimecast
No Tool
Setup required None — works in 30 seconds Weeks of IT deployment & DNS changes
Works on personal email Any inbox, any device Corporate mail only
Explains why it's dangerous Full reasoning + next steps Silent block or generic flag
Live URL sandbox Every submission Enterprise tiers only — extra cost
QR code detection Built-in Add-on module — additional cost
Image & OCR analysis Built-in Limited or not available
Minimum seats None Typically 500+ users
Starting cost Free tier available $50,000+ per year $0 — and no protection
Account required No account, no app Full IT onboarding required

See what's really at the other end

Static analysis can only see what a URL claims to be. PhishEd Intelligence's sandbox opens suspicious URLs in a fully isolated browser, observes the complete execution path, and reports everything — redirect chains, rendered pages, login form detection, and screenshots. The verdict reflects reality, not just metadata.

Redirect chain unwinding

Multi-hop redirects designed to obscure the final destination are followed to the end and fully recorded.

Credential harvesting detection

Login-page clones and password-stealing forms are identified at the rendered page level — not just the URL.

Isolated execution

Every detonation runs in a fresh ephemeral container. No payload can escape to your devices or network.

Screenshot capture

A full-page screenshot of the detonation is captured and stored as evidence, included with every sandbox report.

Three possible outcomes

Every analysis concludes with one of three verdicts, a 0–100 risk score, and a full explanation of the signals that drove the decision.

Likely Phishing

Strong phishing indicators detected — spoofed domains, credential harvesting language, brand impersonation, malicious sandbox findings, or high-pressure deceptive requests. Do not interact.

70–100
Risk score
Suspicious

Concerning signals present but no definitive proof of malicious intent. Proceed with caution — verify the sender through a separate trusted channel before clicking anything.

40–69
Risk score
Likely Safe

No significant phishing indicators found across text, images, links, and sandbox inspection. The message appears legitimate — though no automated system is infallible. Always apply judgment.

0–39
Risk score

Built to handle sensitive content safely

You're forwarding potentially sensitive emails. Here's exactly how we handle them.

Encrypted at rest

Submitted content is stored in encrypted S3 with no public access and strict least-privilege IAM policies throughout the pipeline.

No body logging

Message body content is never written to logs. Only structural metadata — sender, subject, URL list — is recorded for debugging purposes.

Sandboxed execution

URLs are detonated in fully isolated ephemeral containers. No payload can escape to your network or devices — ever.

Rate limited

Analysis is rate-limited per sender to prevent abuse and ensure fair, consistent availability across all beta users.

Prompt-injection resistant

All submitted content is treated as untrusted data — never as instructions. Adversarial payloads embedded in messages are neutralized before reaching the AI models.

Minimal retention

Submitted content is retained only as long as needed for analysis and audit. No data is sold or used to train AI models.

AI-assisted analysis disclaimer: PhishEd Intelligence uses AI models to assess phishing risk. Results are for informational purposes only and do not constitute a guarantee of safety or threat. Always apply your own judgment. PhishEd Intelligence is not responsible for decisions made based on its verdicts. This service is currently in private beta.

Simple, transparent pricing

Start for free — no credit card, no setup. Upgrade when your team needs more.

Free
$0/month

Everything you need to get started. No credit card required.


  • 5 scans per 24 hours
  • Email submission
  • Full two-stage AI analysis
  • Live URL sandbox inspection
  • OCR & QR code detection
  • Verdict, risk score & next steps
  • API access
  • Team management
Join Waitlist
Team
Coming Soon

For organizations protecting their entire staff — zero IT deployment required.


  • Everything in Pro
  • Unlimited team members
  • Admin dashboard
  • Centralized reporting
  • Custom rate limits
  • SSO (SAML / OIDC)
  • Dedicated support
  • SLA guarantee

Common questions

No account, app, or signup will ever be required. Forward any suspicious email from your existing address and you'll receive the verdict reply within seconds. Join the waitlist to be notified when we launch.

Forwarded content is processed automatically by our AI pipeline — no human reads your emails. Message body content is never written to logs. Only structural metadata like sender, subject, and extracted URLs are recorded for debugging purposes.

Submitted content is retained only as long as necessary for analysis and audit purposes. Your data is never sold, shared with third parties, or used to train AI models.

Yes — it works with any email client on any device. Gmail, Outlook, Apple Mail, Proton Mail, Yahoo Mail, and any other app that can forward an email. No plugins, extensions, or integrations required.

Spam filters score emails on bulk sending patterns and known bad domains — they block, but don't explain. PhishEd Intelligence reads embedded images with OCR, decodes QR codes, opens suspicious URLs in a live isolated browser, and tells you exactly why something is dangerous and what to do about it.

PhishEd Intelligence is an AI-assisted tool and results are advisory — not a guarantee. The two-stage AI and live sandbox significantly reduce false positives, but no automated system is perfect. Always apply your own judgment, especially for borderline verdicts marked Suspicious.

Yes. There's nothing to install or configure — just share the analysis address with your team. Each sender gets their own independent rate limit. Business plans with higher limits and centralized reporting are coming soon. Join the waitlist for early access.

Credential phishing, brand impersonation, business email compromise (BEC), smishing via screenshot, QR-based phishing (quishing), image-only evasion attacks, fake invoice fraud, malicious redirect chains, and lookalike domain spoofing.

Join the waitlist

We're wrapping up private beta. Drop your email and we'll let you know the moment we open access — one email, no spam.

Enter your email to be notified at launch:

Limited spots available  ·  No spam  ·  Unsubscribe any time

Ready to stop phishing before it stops you?

Join the private beta — no account required, no app to install. Forward your first suspicious email in under a minute.

No credit card  ·  No setup  ·  No spam