Financial Times Europe - 26.10.2019 - 27.10.2019

26 October/27 October 2019 ★ FTWeekend 19

Spectrum

says Ajder. A week after the video was

released, junior officers attempted a

coupd’état,whichwasquicklycrushed.

In fact, Deeptrace did not find evi-

dence of manipulation but, for Ajder,

that may be irrelevant. What was

important was the uncertainty created.

“Even before these videos become very

goodorverywidespread,wearealready

seeing the spectre of deepfakes haunt-

ing people,” he says. “It really drives

how powerful the mere doubt

is... about any videos we already want

tobefake.”

Kanishk Karan, a researcher at the

Digital Forensic Research Lab, part of

the US think-tank Atlantic Council,

points to another potential deepfake,

this time in Malaysia: a video alleging to

show economic affairs minister zminA

Ali in a trystwith another minister’s

maleaide.

Given Malaysia’s colonial-era laws

and persistent discrimination against

LGBT communities, the footage,

released in June, naturally provoked

controversy.“Alotofpeopleweresaying

it’s a deepfake version of him,” says

Karan. “On the other side, the opposi-

tion is saying that it’s not a deepfake, it’s

a real confession.” To date, the scandal

hasnottoppledAli.

Deepfakes may be particularly

destructive in countries such as India or

Brazil. In both, there is heavy use of

WhatsApp, a platform that lends itself

to videos and images and whose closed

nature also comes with a sense of secu-

rityandtrust.

Bothcountrieshavelargepopulations

without basic literacy, Narayanan of the

Oxford Internet Institute points out,

making it difficult to generate media lit-

eracy. As is often the case with disinfor-

mation, vulnerable populations are

mostatrisk.

“The internet represents the peak of

the fourth industrial revolution. These

are communities that haven’t reaped

thebenefitsofthefirstone—theydonot

have the knowhow to begin to under-

stand that a computer can create this,”

shesays.

Those creating the technology to fight

deepfakes on the ground are divided

into two broad camps. The first is detec-

tion, identifying fake videos and images

as they emerge. Deeptrace is one of the

companies in that space, explains chief

executive Giorgio Patrini, as he calls

from the company’s Amsterdam head-

quarterstodemonstrateitssystem.

The video on my screen looks rather

like an earlier version of Windows

Movie Maker but in navy corporate col-

ours. On the right-hand side are four

videos. Among them is the now-famous

deepfake ofMark Zuckerberg roducedp

for a political art installation and

anotherofRowanAtkinson’sface uper-s

imposedonDonaldTrump’sbody.

The video Patrini drags over to the

left-hand side is unfamiliar, however,

andcomesfromaTaiwanesechannelon

YouTube. It features a young woman

smiling and talking to the camera. Over

the connection, the video comes across

as slightly choppy, but hardly enough to

suggestsomethingisoutoftheordinary.

When he hits play, a red box playing

over her features, flashing percentages,

reveals that it is a fake: a fan wearing a

K-Pop singer’s face. “We’ve seen a cou-

ple of things that are coming out on

Asian markets and are being sold for

pennies on these digital marketplaces,”

says Ajder. This video is harmless, but

Patrini says that (female)K-Pop singers

havebecomemajortargetsoffakeporn.

PatriniexplainsthatDeeptrace’stech-

nology is trained on the thousands of

deepfakes that the company has pieced

together from across the internet. “We

have, to the best of our knowledge, the

largest network of fake videos out

there,”hesays.

The job is a continuous one, though.

“Things a year ago don’t compare,” he

says. “We see thousands of people con-

Continuedfrompage 18

tributing to small tweaks to the technol-

ogyonGitHub,doingitasahobby.”

Farid, the Berkeley professor, is also

working on detection, focusing prima-

rily on public figures, including world

leaders. His system analyses hours of

video of their conversations, including

interviews, speeches and rallies. From

there, it focuses on the specific idiosyn-

crasies of their speech patterns and

expressions.

“When Obama would deliver bad

news, he would frown, his brow would

furrow and he would slightly tilt his

head downwards,” he says. “If he was

being funny he would smile and look up

to the left... Everyone has a different

cadence to how their expressions

change.”

For now, Farid thinks he is one step

ahead of the deepfake producers. “I’m

analysing eyebrows and head move-

ments,” he says. “The GANs don’t know

that there’s a face there.” In the long

term, however, he is pessimistic. “It’s an

arms race and, at the end of the day, we

know we’re going to lose — but we’re

going to take it out of the hands of the

amateur and move it into the hands of

fewerpeople.”

Dr Wael Abd-Almageed, a senior sci-

entistattheUniversityofSouthernCali-

fornia, represents yet another attempt

at detection. His work strings together

several video frames to verify whether

an image is a deepfake. Nevertheless, he

is quick to acknowledge that his

research may unintentionally feed into

improvingfuturedeepfakes.

“My anticipation is the people who

create them will see our paper and try

and improve their methods to fool our

detector,” he says. “If you think deep-

fakes are a problem now, they will be

muchharderinthenextcoupleofyears.”

The second method of combating

deepfakes focuses on improving trust in

videos. Truepic, a San Diego-based

start-up, has been trying to fight manip-

ulated videos and photos for four years,

with experts such as Farid on its advi-

soryboard.

Jeffrey McGregor, Truepic’s chief

executive, says the company launched

in response to a spate of manipulated

pictures online. “Deepfakes will forever

be generated... What Truepic is aim-

ing to do instead of detecting them, is

establishingtruth.”

Truepic has produced a camera app

for everyday usage. “When you tap on

that shutter button, we’re capturing all

of the geospatial data — GPS sensors,

barometricpressures,theheadingofthe

device and securely transforming that

toTruepic’sverificationserver.”

There, the company runs tests to

check whether the image has been

manipulated. If it has not, Truepic

uploads a verified version to its website,

whichcanbesharedwithotherparties.

McGregor says that Truepic has

already found business uses with insur-

ers and lenders. The company is also

working with NGOs who have a particu-

larneedforverifiedimages.“Oneexam-

pleistheSyrianAmericanMedicalSoci-

ety — they’ve used Truepic to document

someoftheeventsthatarehappeningin

Syria,”hesays.

Amber, a San Francisco-based

start-up,producesdetectionsoftwareas

well as Amber Authenticate, a camera

appthatgenerates“hashes”—represen-

tations of the data — that are uploaded

to a public blockchain as users shoot a

video. If the veracity needs to be

checked — for example, in a courtroom

— any differences between hashes can

show whether it has been tampered

with. “Video or audio being used as evi-

dence should not be operating in proba-

bilities,” says Shamir Allibhai, Amber’s

chiefexecutive.

Yet while entrepreneurs and academ-

ics can produce software to fight deep-

fakes,social-mediagiantsmustalsotus-

sle with them. YouTube told the FT that

it was aware of the issue and was work-

ingonit.Thevideoplatformdidremove

thealteredvideoofPelosi.

It remains unclear, however, what

policies it might invoke that could stop

users taking parody videos and repost-

ing them as if they were real, as with the

Renzideepfake.

Of the Big Tech companies, it is Face-

book that has started to take the lead in

looking for technical solutions. In Sep-

tember, it announced the launch of the

Deepfake Detection Challenge along-

side Microsoft, academics in the US and

UK, and an industry consortium called

thePartnershiponAI.

“Better late than never,” says Farid,

who is among the scientists involved.

“Yes, this is good — YouTube and Twit-

tershouldbedoingthistoo—butthere’s

a second part, the policy issue.” He

points to the altered video of Nancy

Pelosi uploaded to Facebook as a prime

example of this dimension. “Facebook

knew it was fake within seconds. They

also said, ‘We are not the arbiters of

truth.’”Thevideostayedup.

In itsrefusal to act, acebook strivedF

to stay within the limits of the Commu-

nicationDecencyAct,Section230—leg-

islation from 1996. It counts websites as

platforms rather than publishers, to

promote free speech, but has come

under increasing criticism for seeming

to enable companies to avoid liability

for the content they host. The result

is an often piecemeal approach to con-

tentissues.

Electoral systemsare also lagging. A

spokesperson for the UK’s Electoral

Commissionsaidthatdeepfakesarejust

onechallengeposedbytheriseofdigital

campaigning. While printed material is

required by law to have imprints show-

ing authorship, this does not apply to

electronic content — a potentially dan-

gerousloophole.

One way to deal with this would be

through enacting clear regulation.

MutaleNkonde,afellowattheBerkman

Klein Center at Harvard University, was

among those involved in helping draft

the Defending Each and Every Person

from False Appearances by Keeping

Exploitation Subject (DEEPFAKES) to

AccountabilityAct.

“It became incredibly important to

enter a piece of legislation,” she says. “As

we move towards 2020, we may be sub-

ject to supposed video evidence and we

need a way of identifying what may look

real [but is not].” She says that there are

fearsthatbothChinaandIrancouldturn

todeepfakesasatooltoattacktheUS.

Yet these dangers have to be dealt

with in the framework of Section 230.

The compromise for Nkonde and her

colleagues was to treat deepfakes as a

consumer-rights issue, making it about

fraudulent representation. The DEEP-

FAKES Accountability Act, referred to

the subcommittee on Crime, Terrorism

and Homeland Security in June, would

make deepfakes for purposes such as

fake porn, disinformation or election

interferenceillegal.

Those synthetic videos produced for

purposes such as parody or education

would need to be watermarked. But

Nkonde says that even as someone who

helped draw up the bill, she now ques-

tionsitsfeasibility.

“The issue with watermarking... is

the technical architecture completely

changes the video,” she says. “It’s a com-

pletely new piece of video.” Trying to

prove something is a fake without

reference to the “real” footage would be

extremelyhard.

She also worries that watermarking

would lead to false positives, or

that canny developers could try to have

real videos flagged as deepfakes. “We

may end up having to actually favour

some type of ban or moratorium until

we get further research in all the differ-

entways[videos]couldbefalsified,”she

suggests. “We’re falling foul to how fast

techismoving.”

While the rate of progress is astound-

ing, experts remain unconvinced about

a deepfake apocalypse in the political

sphere. It is the plausible deniability the

technology offers that remains its great-

est power. Doermann at the University

of Buffalo says that in the US at least,

where public awareness of the technol-

ogy is growing, an extremely high-qual-

ity deepfake would be needed to change

the course of electoral history in

2020. “It would take a massive amount
      of computing power. It’s not going
      to be a rogue person, it would be a
      nationstate.”
      Ajder is also willing to admit that as
      with other advances in AI, scepticism
      around the technology is fair. But he is
      certain, nevertheless, that deepfakes, as
      a form of disinformation, are danger-
      ous. “They appeal to a different kind of
      truth,” he says. “The concept of truth
      has never been as solid as we like to
      think.” Voters who see a video of a poli-
      tician behaving in a way they expect
      them to might understand it is a fake,
      but believe it represents an underlying
      realitytotheircharacter.
      We no longer have the luxury of“
      [deciding] when we can suspend our
      reality,”Ajderconcludes.
      That is evident in the response to the
      altered video of Pelosi. Many comments
      on The Washington Post’s explainer on
      YouTubereflecttheviewthatthespecif-
      ics of the video do not matter. “She still
      sounds drunk and really messed up at
      normal speed... can’t hear no big dif-
      ference,” says one user. That comment
      received564likes.

SiddharthVenkataramakrishnanisanFT

leaderwriter

T

he warning klaxon has

been sounding for years:

artificial intelligence (AI)

will transform the job mar-

ket. The conversation in

the tech world has now turned from try-

ing to convince people of this future to

analysing how and when machines will

enter the workforce independently, and

what that will mean for humans.

A lot of research has been devoted to

pinpointing jobs at risk versus those

considered “safe”, and to defining the

skillsrequired in this new world.

We know that algorithms can aug-

ment human decision-making by find-

ing unexpected correlations in large

data sets. This will allow machines to

displace some jobs that require rote

tasks, assist on other tasks and, ulti-

mately, create some completely new

jobs, such as AI data labellers or algo-

rithm designers.

Fewer studies have been done on how

this might play out in gender terms, but

many futurists are optimistic about

women’s chances of surviving in this

new job market. Why? Some of the

most valued skills in an AI-assisted

workforce will be in areas where robots

fall short: social and emotional intelli-

gence (EI, as opposed to AI).

A report from consultants Capgemini

found that 83 per cent of organisations

surveyed believe an EI workforce will be

a requisite for success in coming years.

If the labour market values empathy,

multitasking, collaboration and com-

passion (attributes historically associ-

ated with women), then women, so the

argument goes, may stand a better

chance of being recruited and retained.

But there is a twist. To flourish in an

AI-driven workforce, a significant

degree of retraining and adaptation will

be required.

The real risks for the female work-

force lie not in the types of jobs they are

suited for, but in whether they can

make smooth job transitions. Will

women be able to reskill and network

their way into new and adjacent occu-

pations as labour demands shift?

A June report from the McKinsey Glo-

bal Institute examined this question in

10 countries, including six developed

economies and four emerging econo-

mies. They found that women and men

face a similar scale of potential job

losses and gains, with an average of 20

per cent of female workers, or 107m

women, potentially being displaced by

automation by 2030, compared

with 21 per cent for men.

But the report also suggested

that up to 160m women worldwide

(and 274m men) may need to tran-

sition across occupations to remain

in work. “If women take advantage

of transition opportunities, they

could maintain their current share

of employment; if they cannot,

gender inequality in work could

worsen,”theauthorswrote.

The adaptation challenges for

womenare multi-faceted — and

not necessarily related to whether

they are suited to more technical

jobs such as algorithm-writing and

data-modelling.

In emerging economies such as

India, for example, new occupations

could require a secondary education,

posing a challenge where female educ-

ation rates continue to lag behind

men’s. Another result of more auto-

mation will be a demand for labour

mobility and flexibility, with workers

required to travel between occupations,

employers, sectors or even countries.

However, women are often less

mobile than men “because they spend

so much more time than men on unpaid

care work — more than 1.1 trillion hours

a year, compared with under 400bn

hours for men,” the report said.

In addition, women often find them-

selves excluded from leadership net-

works that allow men to enhance their

skills, find mentors and identify

job opportunities. This is some-

times the result of caring responsi-

bilities: collecting a child from

nursery or visiting an elderly par-

ent may be prioritised over post-

work drinks in the pub.

But it also reflects an uncon-

scious bias that still causes many

people to associate leadership

traits with men.

So while we can celebrate the

notion that the age of machines

may put a value on so-called “soft

skills” such as empathy and com-

munication, we will need to con-

tinue to focus on the banal practi-

calities of work in 2030 if women

are to be included.

Ultimately, it comes back to the same

challenges we have today: will women

have access to the same networks, edu-

cation and opportunities that their male

counterparts do? If not, an AI world

won’t look so different after all.

MadhumitaMurgiaistheFT’s

Europeantechnologycorrespondent

When networks rule

ech WorldT Could women be better placed to

thrive in a workforce transformed by AI? Or, asks

Madhumita Murgia, will the old biases still apply?

âtéP

‘There are communities

that haven’t (even) reaped

the benefits of the first

industrial revolution.

They do not have the

know-how to begin to

understand that a

computer can create this’

Vidya Narayanan, researcher,

Oxford Internet Institute

From top: a deepfake

video of former Italian

prime minister Metteo

Renzi, broadcast on a

satirical TV show last

month; created as a

political art installation,

this deepfake of Facebook

CEO Mark Zuckerberg

went viral when it was

posted in June; in this

videothe face of British

comedian Rowan Atkinson

is superimposed on

Donald Trump’s body

ouTubeY

Getty Images

OCTOBER 26 2019 Section:Weekend Time: 25/10/2019- 15:47 User:adrian.justins Page Name:WIN19, Part,Page,Edition:WIN, 19, 1