AI can rephrase anything — which is why AI detectors and plagiarism scores are losing the fight. PowerNotes makes integrity visible instead: every source captured, every citation automatic, every AI conversation on the record. When the process is this clear, there's nothing left to detect.
Trusted at 65+ schools and universities worldwide · Two-semester pilot programs available for IB schools and institutions across Asia
Plagiarism detection was built for a world where cheating meant copying — and copying left fingerprints a database could match. That world is gone. Generative AI rephrases instantly, endlessly, and originally: text that was never written anywhere before will never match a repository of past papers. The dishonest student no longer copies. And the detectors built to respond — AI-probability scores — have produced a different casualty: honest students flagged by statistical guesswork, with no way to prove their innocence.
What's actually left of the old plagiarism problem is the accidental kind: the quote that lost its citation, the paraphrase that drifted too close to its source, the reference list rebuilt from memory at 2 a.m. That problem is real — and it isn't solved by scoring the finished paper. It's solved at the moment the source enters the student's work.
PowerNotes captures attribution automatically, while students research — so citations can't go missing, because they were never manual in the first place.
Students highlight anything — a web page, a library database article, a PDF — and PowerNotes saves the passage, the source, and a formatted citation (APA, MLA, Chicago and more) in one action. The bibliography builds itself as the research happens.
Highlights and notes remain permanently linked to the exact source passage they came from. When a quote lands in the outline — and later in the paper — its origin travels with it. Nothing gets orphaned in a copy-paste doc.
When a paper is submitted, PowerNotes checks its claims against the student's own research record. Statements backed by sources are shown as cited; statements with no trail are flagged as unsupported — with the reasoning visible — so missing attribution gets fixed as feedback, not punished as misconduct.
A finished document can be faked. Months of research can't. PowerNotes records the entire arc of student work — sources found, passages highlighted, notes taken, outlines built, AI conversations held — timestamped as it happens. If a submitted paper doesn't match the process behind it, that mismatch is immediately visible. No probability score. No accusation. Just evidence anyone can inspect — including the student.
Every project carries a defensible, timestamped record of the student's own thinking. For students, it's protection — proof of process when questions arise. For teachers, it's the answer to "did they do the work?" without reading surveillance reports.
Students get real AI — leading models from Anthropic, OpenAI, Google and more — grounded in their own sources, inside guardrails each teacher sets per assignment in plain language. Every prompt and response is logged. Allow brainstorming, block drafting, require Socratic dialogue: your policy, enforced automatically.
Work generated outside the platform shows up as writing with no research behind it — no sources, no notes, no development, claims with no trail. Faking an entire research process takes longer than doing one. That's the point: PowerNotes makes honest work the fast path.
Schools aren't abandoning integrity — they're abandoning a method. Since 2023, more than fifty universities across five countries — including Vanderbilt, Yale, and Johns Hopkins — have disabled AI writing detection, citing false positives that a probability model can't eliminate. Independent research has shown detection rates collapsing to single digits once AI text is lightly reworded. In 2026, a US federal court sided with a student whose paper was wrongly flagged as 100% AI. And surveys show wrongful AI accusations fall hardest on non-native English speakers and students of color.
The verdict from institutions themselves: detection scores can start a conversation, but they can't carry a case. Process evidence can. Even the detection industry now agrees — its newest products record process instead of guessing at authorship. The difference is where that process lives: in a separate, monitored writing box — or in the research workspace students already choose to use.
The same integrity questions, answered two different ways.
| Detection-era tools | PowerNotes | |
|---|---|---|
| What it examines | The finished document, after the fact | The whole research process, as it happens |
| How it judges AI use | A statistical probability score students can't appeal | A complete log of every AI prompt, response, and rule |
| Accidental plagiarism | Flagged at submission — often as suspected misconduct | Prevented at capture — citations are automatic from the first highlight |
| Deliberate cheating | Rephrased AI text routinely slips through | Output must match a visible process; work with no trail exposes itself |
| Honest students | Carry the burden of proving innocence | Hold timestamped proof of their own work |
| Student AI access | Not included — AI is treated as a threat to detect | Included — governed, grounded, budgeted AI for every student |
| Where students work | A monitored writing environment | Their own browser, library, and word processor |
| Cost structure | Base license + AI-detection add-on + process add-on | One platform; AI metered against a budget you cap |
The IB's approach to academic integrity has always been about authenticity of process — not detection scores. That's exactly what PowerNotes produces. For the Extended Essay and Internal Assessments, students build a live record of sources, notes, and reasoning from the first day of research; supervisors see development over months, not a single upload at the deadline; and citations in every major style are generated the moment a source is used — in any language your students research in.
PowerNotes runs structured two-semester pilots with universities and schools across Asia — currently in Taiwan and Malaysia, with regional support on the ground through our partner iGroup and its member companies. Pilots include faculty training and workshops, LMS integration (Moodle, Canvas and other LTI 1.3 systems), library search integration with EBSCO EDS and Ex Libris Primo/Summon, and usage data your institution can learn from. Student data is handled to GDPR and PDPA standards, students own their own work, and school data is never used to train AI models.
See what visible process looks like — for one class or a whole campus. Demos take 45 minutes; two-semester pilots include training, integration, and usage data your institution keeps.
