Topics
belated
AI
Amazon
Image Credits:ForceField
Apps
Biotech & Health
Climate
Image Credits:ForceField
Cloud Computing
Commerce
Crypto
go-ahead
EVs
Fintech
fundraise
Gadgets
Gaming
Government & Policy
Hardware
Layoffs
Media & Entertainment
Meta
Microsoft
privateness
Robotics
Security
Social
Space
Startups
TikTok
Transportation
Venture
More from TechCrunch
event
Startup Battlefield
StrictlyVC
Podcasts
television
Partner Content
TechCrunch Brand Studio
Crunchboard
Contact Us
A Modern inauguration setting out to combat the scourge of deepfakes andspoofed evidencein the age of AI is showing off its wares on the Startup Battlefield degree atTechCrunch Disrupt 2024this week . ForceFieldis build a set of “ patent of invention - pending ” APIs dubbed MARQ , the first of which is designed to authenticate content .
For circumstance , a 2022reportfrom European law enforcement agency Europol predicted that 90 % of all on-line content will be AI - get by 2026 . The internet is already inundated with manipulated capacity — either totally machine - generated , or made with the help of AI — forcing bighearted tech company to develop novel tools to treat the job . Google last monthannounced that it will soonbegin flagging AI - yield images in its hunting railway locomotive , follow in the footsteps ofMeta , which was already doing soon Facebook and Instagram — though withquestionable truth .
YouTube , meanwhile , now allows people torequest the removal of contentthat simulates their face or representative .
The issue of “ fake ” content is n’t run off , and it ’s a problem that has given rise to a swathe of startup that are tackling the trouble through various means , such as Clarity , Reality Defender , andTruepic , which have all raise sizeable sums from braggy - name backers in recent years .
ForceField , for its part , is concenter on the machine stage rather than the content — this meansanydata stream , on any gimmick , where ForceField ’s technology is present . This could be a wandering app with ForceField ’s API incorporate , or a surveillance tv camera ’s telecasting - management system or drone ’s hardware — it can “ sign and hash for confirmation ” in real - time .
‘Chain of custody’
ForceField ’s first production , calledHashMarq , verifies substance submissions and labels it as authentic , replete with a badge of genuineness . Rather than using tangential techniques such asadversarial AI(i.e . fighting AI with AI ) orwatermarking , Forcefield ’s engineering establishes a “ chain of custody ” to found provenience by compute out whether a specific piece of cognitive content has been altered — it does so by face at metadata from data streams and affiliated devices . What we ’re blab about here is blockchain engineering , though ForceField founder and CEOMC Spano(pictured above ) try that this is n’t a “ web3 ” inauguration .
“ We ’re leverage blockchain , we ’re not using coin or crypto — we ’re not aweb3company , ” Spano told TechCrunch . “ We ’re using blockchain in the most innate and purposeful way , which is voguish catching . We authenticate anything collected , like a picture , picture , an audio file , a screenshot on a mobile machine — but we also secure and authenticate any multimedia system collected on any IoT , which means anything from a BlackBerry to a robot . But any streaming information as it ’s collected , we ’re signing it , we ’re hash it , and then we ’re using ledger engineering . ”
Join us at TechCrunch Sessions: AI
Exhibit at TechCrunch Sessions: AI
These “ smart contract bridge ” inhabit on the blockchain ( though ForceField does n’t reveal which one ) to ensure their certificate , transparency , and immutableness — they ca n’t be changed . endeavour client — which may let in anyone from principal data officers to swear and guard team — can reach the technology through an API or SDK which they desegregate into their own applications . So HashMarq basically wait on as a digital certificate for message that has happen its proof hinderance .
“ We ’re building tech for the expert guys — we ’re previse that the market , meaning governments , go-ahead , small companies , even individuals , will necessitate to testify what they ’re set up out in the world is genuine , ” Spano said . “ This is liken to something like what Meta and other companies are doing , where they ’re fundamentally mark something that ’s generated by AI . ”
So what , exactly , does ForceField ’s technology look at when assessing the veracity of a give way asset ?
“ We ’re look at about 90 proof point , but I ca n’t share all of them , ” Spano sound out . “ The feeling is that we ’re looking at the geospatial intelligence service around that machine , what was around at the time , what signals , what other machine , and the time . What time was this gadget , specifically , collecting this video ? And how long was it from the ingathering to the submission ? Those period alone can tell journalists or news show system or law of nature enforcement so much more than what they would originally have , and it also can nail a geographical position . ”
Though ForceField formally occur to fruition in 2021 , Spano says the genesis for the product came in 2018 when she was assault in New York City , and after face major obstacles both in gathering and submitting the grounds required by police force enforcement .
“ I knew that there were cameras , I knew that the grounds was able to be gathered , ” Spano said . “ But when I shell out with NYPD to get the evidence , they would n’t get a subpoena — so I had to go to the owner of the camera myself to request the footage . I record it with my phone , and when I turned that transcription into NYPD , they say it was n’t admissible . And that ’s when I realized the justice system commence way earlier than in the lawcourt . ”
And this gets to the Crux Australis of the problem that Spano and ForceField are trying to solve — a problem that has become even greater in the age of reproductive AI , where anyone with a laptop is capable to make plausible , but entirely fake , “ evidence . ”
Proof-of-concept
It is still other Clarence Day , but ForceField is developing a validation - of - construct ( PoC ) withinsurance giant Erie , while Spano says the company also has missive of design signed with three extra client , with plan to finalize the PoC and launch commercially in Q1 2025 .
While insurance policy will be a big market for ForceField at first , Spano says that with the advent of AI and the power for moderately much anyone to create authentic - looking assets , there ’s no real limits on the industries in which it can utilise to . A citizen diary keeper , for example , who has captured a television on their mobile twist , could submit it to a newsworthiness post , which would integrate with ForceField ’s API to find out the authenticity of the source data . law of nature enforcement , too , could employ the technology to verify informant or victim submission .
In terms of financing , the company has so far been fund via a mixture of bootstrapping ( Spano sold a previous society ) and capital letter provided by angels , including actor Debra Messing , as well asWilfredo Fernandez , X ’s head of government affairs and a scout at VC firm Lightspeed . ForceField has also received money through itsparticipation in Techstarslast year , while it ’s just taken on its first institutional financing via a strategic investiture from its customer , Erie Insurance .
Spano also aver that she skip to close up a germ round by the end of the year , add together that she believes one of the most significant facets of the society is that it is an entirely woman - led caller in a very male - command space .
“ There ’s a really big issue with funding solo women teams here in the U.S. , ” Spano say . “ I am the only woman building this type of technology decent now , every other competitor are all led by men — so having a fair sex ’s voice , and foil , and money plant , and true statement … is really important . ”
