Industries / Retail, hospitality and personal services / Beauty salons and spas
Hair
Cutting, coloring, waving, straightening, braiding and barbering hair for paying clients in a salon or barbershop.
5
Milestones achieved, from 1909 to 2023.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1909
Waves set with heat
Charles Nessler, trading as C. Nestle & Co., filed a British patent on 6 February 1909 for waving hair on the head: each lock was wound on a curler, wrapped in moist flannel, soaked in lotion and heated from outside. Electric permanent waving machines followed in salons; the Science Museum Group holds examples from about 1930 and about 1936.
patents.google.comcollection.sciencemuseumgroup.org.ukcollection.sciencemuseumgroup.org.uk
Credit
Charles (Karl) Nessler
1960s
Cuts that need no setting
Vidal Sassoon's geometric cuts, such as the bob that the designer Mary Quant made famous, were cut to fall into shape and needed little styling. The Associated Press called them "wash-and-wear cuts" that freed women "from endless teasing and hairspray".
Credit
Vidal Sassoon, London
2010
Formaldehyde-free labels tested
Oregon OSHA, the state's workplace safety agency, tested 105 samples of keratin hair-smoothing products from 54 salons and on 29 October 2010 reported significant formaldehyde in products labeled "formaldehyde free". On 22 August 2011 the US Food and Drug Administration (FDA) warned the maker of Brazilian Blowout that its product was adulterated and misbranded, and the federal research agency NIOSH measured 11% formaldehyde in one such product and air levels up to 1.3 parts per million in an Ohio salon, against its ceiling of 0.1.
LessonA label on a salon product is a claim until someone tests the bottle.
Credit
Oregon OSHA; US FDA; NIOSH
2022
Straighteners and uterine cancer
The US National Institute of Environmental Health Sciences followed 33,497 women in its Sister Study for almost 11 years. By the institute's estimate, 4.05% of women who used chemical hair straighteners more than four times in the previous year would develop uterine cancer by age 70, against 1.64% of women who never used them. The lead author, Alexandra White, described it as, to the team's knowledge, the first epidemiologic study of the question, and the study did not record brands or ingredients.
Credit
Che-Jung Chang, Alexandra White and colleagues, NIEHS
2023
A swap that hurt kidneys
Doctors at 14 medical centers in Israel described 26 people with severe acute kidney injury after salon hair-straightening treatments between 2019 and 2022; three needed temporary dialysis. Most straighteners sold there used glycolic acid derivatives, and kidney biopsies showed calcium oxalate deposits; a Geneva team reported a similar case in 2024 and pointed to glyoxylic acid in the straightening cream.
LessonThe ingredient that replaces a suspect one needs its own safety record.
Credit
A. Bnaya and colleagues; University Hospitals of Geneva
Watch for
FDA formaldehyde ban
The FDA says its proposed rule to ban formaldehyde and formaldehyde-releasing chemicals in hair-smoothing and hair-straightening products "continues to remain a priority". Its target date has moved from October 2023 to April 2024, November 2024, March 2025, July 2025 and December 2025, according to the Environmental Working Group, and the rule was still unpublished in January 2026.
Achieved whena final rule is published in the Federal Register.
Claimed by
US Food and Drug Administration, January 2026
A braiding assistant
HaloBraid says a stylist starts a knotless or box braid by hand and its device finishes the rest "in seconds". The company raised $7 million in June 2026 and plans to place the device in salons later that year.
Achieved whensalons outside the company use it and an independent party publishes braid times and hair condition against braiding by hand.
Claimed by
HaloBraid, founder Yinka Ogunbiyi, June 2026
Open
01
Test formaldehyde-free claims
A public table of salon hair-smoothing and straightening products sold today as "formaldehyde-free", each tested in an accredited lab for free formaldehyde and methylene glycol (formaldehyde dissolved in water), and for glyoxylic acid, next to what its label and safety data sheet say. The bottles are tested as sold; nothing is heated or used on anyone.
The briefCredit · unclaimed02
Open hair color results
An open dataset of hair color outcomes: swatches of human hair at known starting shades, each colored with a recorded formula and measured with a color meter before and after, plus a simple model that predicts the result from the starting shade and formula. The work is done on cut swatches with gloves in a ventilated room, never on a person.
The briefCredit · unclaimed
Open 01 · the brief
Test formaldehyde-free claims
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public table of salon hair-smoothing and straightening products sold today as "formaldehyde-free", each tested in an accredited lab for free formaldehyde and methylene glycol (formaldehyde dissolved in water), and for glyoxylic acid, next to what its label and safety data sheet say. The bottles are tested as sold; nothing is heated or used on anyone.
Why it is still open
Regulators still find formaldehyde one product at a time: in March 2026 UK trading standards recalled a Brazilian keratin treatment sold for professional use because it contained formaldehyde, which is banned in cosmetics there. The broad tests that exposed false labels were run by US agencies in 2010 and 2011, a December 2024 dermatology editorial still relies on product tests published in 2016 and 2020, and a March 2026 review by Toxic-Free Future found formaldehyde releasers listed on 68 of 577 textured hair products sold by Sally Beauty, but it read labels only. The FDA notes that products sold only to professionals need not list ingredients at all.
Built with what exists
Products bought from salon suppliers and online shops; a contract cosmetics lab accredited to ISO/IEC 17025 running a liquid chromatography test for formaldehyde; label photos and safety data sheets; a public spreadsheet.
First step
List 30 products on sale in the US, the EU and Brazil that carry a "formaldehyde-free" or "no formaldehyde" claim, and get price quotes from two accredited labs.
Done looks like
30 products from at least three countries tested, with the lab report, method and label photo for each published, every mismatch filed with the FDA's cosmetic complaint system or the national regulator, and a repeat round after Washington state's limit on formaldehyde releasers starts on 1 January 2027.
Open 02 · the brief
Open hair color results
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open dataset of hair color outcomes: swatches of human hair at known starting shades, each colored with a recorded formula and measured with a color meter before and after, plus a simple model that predicts the result from the starting shade and formula. The work is done on cut swatches with gloves in a ventilated room, never on a person.
Why it is still open
L'Oréal's Coloright machine, shown in January 2022, measures a client's hair color, gray share, length and density and dispenses a formula, but its data and accuracy have not been published. Color makers have patented ways to predict a dye result from the starting color, but searches of GitHub, Hugging Face and the research literature found no open before-and-after color data.
Built with what exists
Human hair swatches sold for lab testing; oxidative hair colors and developers from salon suppliers; a handheld spectrophotometer such as the Nix Spectro 2 reading CIELAB color values; Python with scikit-learn.
First step
Buy 50 swatches across five starting levels, color them with ten shades at one developer strength, and measure each three times.
Done looks like
At least 500 swatch-and-formula pairs released under an open license, and a model that predicts the after color within a color difference (CIEDE2000) of 3 on 100 swatches it was not trained on.