Brave, Fingerprinting and Privacy on the Web Me (the early years) - - PowerPoint PPT Presentation

▶
brave fingerprinting and privacy on the web me the early
SMART_READER_LITE
LIVE PREVIEW

Brave, Fingerprinting and Privacy on the Web Me (the early years) - - PowerPoint PPT Presentation

Jul 01, 2023 38 likes •1.09k views

Brave, Fingerprinting and Privacy on the Web Me (the early years) Grew up in Chicago actual Chicago Law school, then freelance web design Started: Anchorage, AK Ended: Judge Judy Show invitation PhD in Computer Science

slide-1
SLIDE 1

Brave, Fingerprinting and Privacy on the Web

slide-2
SLIDE 2

Me (the early years)

  • Grew up in Chicago


…actual Chicago

  • Law school, then freelance web design


Started: Anchorage, AK 
 Ended: Judge Judy Show invitation

  • PhD in Computer Science


University of Illinois at Chicago

slide-3
SLIDE 3

Me, now

  • Privacy Researcher at Brave


Research to improve privacy in the browser

  • Co-Chair of PING


Privacy reviews of new web standards

  • Academic Collaborator


“Pure” research

slide-4
SLIDE 4

Brave in a Slide

  • Privacy focused
  • Alternative funding model for

the web

  • Research and engineering

focused

  • Browsers and infrastructure

now, more to come…

slide-5
SLIDE 5

Overview

  • 1. Why websites track (and how much)
  • 2. “Classic” tracking
  • 3. Fingerprinting / “passive tracking”
  • 4. Fingerprinting counter measures
  • 5. Anti-finger printing exercise
  • 6. Privacy protections in Brave
  • 7. Wrapping up
slide-6
SLIDE 6

Overview

  • 1. Why websites track (and how much)
  • 2. “Classic” tracking
  • 3. Fingerprinting / “passive tracking”
  • 4. Fingerprinting counter measures
  • 5. Anti-finger printing exercise
  • 6. Privacy protections in Brave
  • 7. Wrapping up
slide-7
SLIDE 7

Why Does Tracking Exist?

slide-8
SLIDE 8
slide-9
SLIDE 9
slide-10
SLIDE 10
slide-11
SLIDE 11
slide-12
SLIDE 12
slide-13
SLIDE 13
slide-14
SLIDE 14
slide-15
SLIDE 15

$ $ $ ¢ ¢ ¢

slide-16
SLIDE 16

$

slide-17
SLIDE 17

Identify “expensive”
 people here Pay a little to advertise
 to them here

slide-18
SLIDE 18
slide-19
SLIDE 19
slide-20
SLIDE 20

But how much…

slide-21
SLIDE 21
slide-22
SLIDE 22
slide-23
SLIDE 23
slide-24
SLIDE 24
slide-25
SLIDE 25

But how much… a lot / too much

slide-26
SLIDE 26

Overview

  • 1. Why websites track
  • 2. “Classic” tracking
  • 3. Fingerprinting / “passive tracking”
  • 4. Fingerprinting counter measures
  • 5. Anti-finger printing exercise
  • 6. Privacy protections in Brave
  • 7. Wrapping up
slide-27
SLIDE 27
slide-28
SLIDE 28
  • Javascript
  • DOM / Initial Web API
  • Netscape
  • Firefox
  • Brave + BAT
slide-29
SLIDE 29

Web 0.0

good-site.com

slide-30
SLIDE 30

Web 0.0

good-site.com GET /home.html <html>…</html>

slide-31
SLIDE 31

Web 0.0

good-site.com GET /home.html <html>…</html> GET /other.html <html>…</html>

slide-32
SLIDE 32

Web 0.0

good-site.com GET /home.html <html>…</html> GET /other.html <html>…</html>

slide-33
SLIDE 33

Birth of the Tracking

  • Problem
  • Authentication?
  • Can’t log in every time
  • HTTP auth is terrible and limited
  • Solution
  • Server gives token to user
  • User returns it on requests
  • Aka “cookies”
slide-34
SLIDE 34

Web 0.0

good-site.com

slide-35
SLIDE 35

Web 0.0

good-site.com GET /home.html <html> + id=XYZ

slide-36
SLIDE 36

Web 0.0

good-site.com GET /home.html <html> + id=XYZ GET /secret.html + id=XYZ <html>…

slide-37
SLIDE 37

Web 0.0

good-site.com GET /home.html <html> + id=XYZ GET /secret.html + id=XYZ <html>… GET /secret.html 🙆🙆

slide-38
SLIDE 38

But in the meantime…

slide-39
SLIDE 39

cat-cuties.com kozy-kittens.com

slide-40
SLIDE 40

cat-cuties.com kozy-kittens.com

slide-41
SLIDE 41

cat-cuties.com kozy-kittens.com

slide-42
SLIDE 42

cat-cuties.com kozy-kittens.com cookies, cookies everywhere…

slide-43
SLIDE 43

Cookies + 3p Resources ———————— Tracking

slide-44
SLIDE 44

44

Site A

slide-45
SLIDE 45

45

Site A Id=abc Tracking Site

slide-46
SLIDE 46

46

Site A Tracking Site Site B

slide-47
SLIDE 47

47

Site A Tracking Site Site B Id=abc

slide-48
SLIDE 48

48

Site A Tracking Site Site B Tracker knows the same person visited A + B

slide-49
SLIDE 49

Tracking Patient Zero

  • The internets “original sin”

  • cross origin resources

  • 3p cookies

  • or both…
  • “I invitent Javascript and


3p script, and I’ve been 
 making up for it ever sense…”
 (paraphrase)

slide-50
SLIDE 50

“Ever-Cookies”

  • Some browsers started fighting back


Brave, Safari, Firefox, extensions…


  • Trackers fought back


Moving IDs information out of cookies, to other location


  • Long list of locations

○ Local and Session Storage ○ HSTS ○ Cache (etags, Cache API, etc) ○ Plugins ○ many many many more…

slide-51
SLIDE 51

Overview

  • 1. Why websites track (and how much)
  • 2. “Classic” tracking
  • 3. Fingerprinting / “passive tracking”
  • 4. Fingerprinting in web standards
  • 5. Fingerprinting counter measures
  • 6. Anti-finger printing exercise
  • 7. Wrapping up
slide-52
SLIDE 52

Fingerprinting, what’s diff?

  • Classic tracking

  • Website stores an id on the client

  • The client returns the id to the server (cookie or JS)

  • The id is what allows re-identification
  • Fingerprinting / passive tracking

  • Website finds things different about each visitor

  • That difference allows re-identification
slide-53
SLIDE 53

Fingerprinting, how

  • Large number of semi-identifiers

  • Browser size

  • Extra fonts

  • Audio hardware

  • Video hardware

  • Installed plugins

  • Color depth

  • etc etc etc…
slide-54
SLIDE 54

All browser users

slide-55
SLIDE 55

All browser users:
 3 billion people You
 1 person in 3 billion

slide-56
SLIDE 56

Firefox
 Users All browser users:
 3 billion people

slide-57
SLIDE 57

Windows users All browser users:
 3 billion people

slide-58
SLIDE 58

Office Fonts All browser users:
 3 billion people

slide-59
SLIDE 59

Sending DNT header All browser users:
 3 billion people

slide-60
SLIDE 60

Using ad blocker All browser users:
 3 billion people

slide-61
SLIDE 61

All browser users:
 3 billion people You
 1 person in 100

slide-62
SLIDE 62

Succeeding at Fingerprinting

  • 1. Breath of fingerprints


Large number of semi-identifiers

  • 2. Depth of fingerprints


How uniquely each identifier can… identify

slide-63
SLIDE 63

Breath (examples)

  • User agent string
  • Installed fonts
  • Canvas / WebGL
  • Hardware (many)
  • Height / width
slide-64
SLIDE 64

User Agent String

  • History of the Browser user-agent string


https://webaim.org/blog/user-agent-string-history/

  • Katamari-Damacy of identifiers
  • Brave / Chrome

Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_0) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/78.0.3904.50 Safari/537.36

  • Easy to extract

  • navigator.userAgent

  • User-Agent:
slide-65
SLIDE 65
  • Three categories of fonts

  • System

  • Local

  • Web
  • “Local” is the tricky part

  • Office

  • Photoshop

  • Goofery
  • Easy to extract

  • plugins

  • css + span + width

Installed Fonts

slide-66
SLIDE 66

[‘Andale Mono', 'Arial', 'Arial Black', 'Arial Hebrew',
 'Arial MT', 'Arial Narrow', 'Arial Rounded MT Bold’…]

slide-67
SLIDE 67

[‘Andale Mono', 'Arial', 'Arial Black', 'Arial Hebrew',
 'Arial MT', 'Arial Narrow', 'Arial Rounded MT Bold’…] <span>Example</span>

slide-68
SLIDE 68

[‘Andale Mono', 'Arial', 'Arial Black', 'Arial Hebrew',
 'Arial MT', 'Arial Narrow', 'Arial Rounded MT Bold’…] <span>Example</span>

Fingerprinter

For each font…

slide-69
SLIDE 69

[‘Andale Mono', 'Arial', 'Arial Black', 'Arial Hebrew',
 'Arial MT', 'Arial Narrow', 'Arial Rounded MT Bold’…] <span>Example</span>

Fingerprinter

For each font… for (const fontName of fonts) { // 1. Apply font to span // 2. Measure width of span // 3. If it changes, user has font… }

slide-70
SLIDE 70

Canvas / WebGL

  • Pixel Perfect: Fingerprinting Canvas in HTML5


Keaton Mowery and Hovav Shacham

  • Drawling APIs


e.g. Drawing lines / shapes

  • Standardized, but subtle differences
  • Easy to extract

  • Create canvas

  • Do some drawing

  • toDataURL()

slide-71
SLIDE 71
slide-72
SLIDE 72

Hovav Shacham The Geometry of Innocent Flesh on the Bone:
 Return-into-libc without Function Calls

slide-73
SLIDE 73

Hardware Identifiers

  • Many Web APIs leak capabilities 

  • number of cores (HTML)

  • number of audio channels (Web Audio API)

  • num shaders and similar (WebGL API)

  • device memory (Device Memory API)

  • network (WebRTC, Network status API)
  • Semi identifying
  • Easy to extract

  • All browsers have subset of the above

  • Most platforms have no permissions
slide-74
SLIDE 74

Height / Width

?

slide-75
SLIDE 75

Height / Width

  • What does it mean?
  • How to extract w/ JavaScript?
  • How to extract w/o JavaScript?
  • Brb 5 min (Go go go go go!)
slide-76
SLIDE 76

Fingerprinting Depth

https://panopticlick.eff.org/

slide-77
SLIDE 77

Fingerprinting in Practice

  • Needs to be in a database…
  • Hash each endpoint
  • Hash each value into a single identifier…
  • Nice implication: “poisionability”…
slide-78
SLIDE 78

Exercise

  • Read fingerprint2.js
  • List as many finger-printing approaches as possible
  • Understand how they’re carried out
  • Predict which are most identifying
slide-79
SLIDE 79

Overview

  • 1. Why websites track (and how much)
  • 2. “Classic” tracking
  • 3. Fingerprinting / “passive tracking”
  • 4. Fingerprinting counter measures
  • 5. Anti-finger printing exercise
  • 6. Privacy protections in Brave
  • 7. Wrapping up
slide-80
SLIDE 80

Fingerprinting Countermeasures

  • Remove the functionality
  • Make the functionality consistent
  • Restrict access (permissions, 1p vs 3p, user gesture, etc)
  • Noise
  • “Privacy budget”
slide-81
SLIDE 81

Remove the functionality

  • Delete JS end point
  • Remove the HTTP header
  • Remove the runtime capability
slide-82
SLIDE 82

Consistency

  • Make every browser return the same value
  • … or, most?
  • Not that diff in practice from “removing”
slide-83
SLIDE 83

Restrict access

  • Permission prompt
  • User gesture
  • 1p vs. 3p
  • “Site engagement”
slide-84
SLIDE 84

Noise

  • Stenography
  • Make different every time
slide-85
SLIDE 85

Privacy Budget

  • Allow some identifiability
  • After “identifiability budget” is exhausted do… something
  • Google folks love it
  • Everyone else is… skeptical
slide-86
SLIDE 86

Overview

  • 1. Why websites track (and how much)
  • 2. “Classic” tracking
  • 3. Fingerprinting / “passive tracking”
  • 4. Fingerprinting counter measures
  • 5. Anti-finger printing exercise
  • 6. Privacy protections in Brave
  • 7. Wrapping up
slide-87
SLIDE 87

Fingerprint2 Again…

  • Choose two fingerprinting vectors to combat
  • Propose counter measures
  • Choose two fingerprinting vectors that are hard
  • Why are counter measures hard?
slide-88
SLIDE 88

Fingerprint2 pt 3…

  • Pretend your the attacker
  • How would you respond to those defenses…
slide-89
SLIDE 89

Fingerprint2 pt 4…

  • Pretend your the defender again
  • How would you modify your defenses given the previous

round…

slide-90
SLIDE 90

Overview

  • 1. Why websites track (and how much)
  • 2. “Classic” tracking
  • 3. Fingerprinting / “passive tracking”
  • 4. Fingerprinting counter measures
  • 5. Anti-finger printing exercise
  • 6. Privacy protections in Brave
  • 7. Wrapping up
slide-91
SLIDE 91

91

Brave Privacy Protections

slide-92
SLIDE 92
  • Shields

  • Global protection

from tracking


  • On by default

  • Can be disabled if

needed

Brave Privacy Protections

slide-93
SLIDE 93

Brave Privacy Protections

  • Block cross site trackers

  • Lists of known tracking

websites


  • Refuse to load

  • Both community and

Brave generated

slide-94
SLIDE 94

94

Blocking Cross-Site Trackers in Brave

  • EasyList and EasyPrivacy


Used by AdBlock Plus, etc.


  • Disconnect


Used by Firefox, extensions


  • uBlock Origin


Excellent blocking extension


  • Brave generated


Open source, shared with community

slide-95
SLIDE 95

Brave Privacy Protections

  • Don’t send

identifiers to third party sites


  • Send to “main” site

  • Same with other

storage methods

slide-96
SLIDE 96

Brave Blocks Tracking Cookies

Site A Id=abc Tracking Site

slide-97
SLIDE 97

Brave Blocks Tracking Cookies

Site A Id=abc Tracking Site

slide-98
SLIDE 98

Brave Blocks Tracking Cookies

Site A Tracking Site Site B

slide-99
SLIDE 99

Brave Blocks Tracking Cookies

Site A Tracking Site Site B Id=abc

slide-100
SLIDE 100

Brave Blocks Tracking Cookies

Site A Tracking Site Site B Tracker can’t link A and B

slide-101
SLIDE 101

Brave Privacy Protections

  • Reduce finger printing

vectors


  • Currently:

○ Hardware identifiers ○ Canvas ○ WebGL ○ Audio

  • Planned:

○ Fonts ○ User agent ○ Screen size

slide-102
SLIDE 102
  • Restrictions on third-party

scripts


  • Identifying tracking behaviors,

not just scripts / URLs


  • Query parameters filtering

  • Bounce tracking

  • Much more…

Under Exploration Possible Privacy Protections

slide-103
SLIDE 103

Overview

  • 1. Why websites track (and how much)
  • 2. “Classic” tracking
  • 3. Fingerprinting / “passive tracking”
  • 4. Fingerprinting counter measures
  • 5. Anti-finger printing exercise
  • 6. Privacy protections in Brave
  • 7. Wrapping up
slide-104
SLIDE 104

Unasked for Advice

  • Brave is hiring, keep us in mind
  • Privacy is more than just web, there’s lots to do
  • Don’t accept privacy as a feature…
  • Choose your employer with values in mind
slide-105
SLIDE 105

Thanks!

  • Pete Snyder


Privacy Researcher
 pes@brave.com
 @pes10k

  • Questions?

  • Standards work?

  • Privacy jobs?

  • Brave business model

  • BAT / Block chain

  • Anything else?